Home

S-PLUS User's Guide

1. el Lf FF 33 ast mia bea dimde Eti store Etil store Eti store ri bar s ri bar o kei bxp s ti bxp bxp c model object is shown in figure 14 1 Jox Match jan Contents of llast Im Page 1 of fi Object Pos Data Class Inheritance 1 call EEE ms 1 terms terms Sane ntrasts 1 list FESE residual 1 double Delete sign 1 it Summary k 1 integer Plot 1 double upper Predio ects 1 double 15 ed values 1 double COGE B 1 double Help efficients 1 double Figure 14 1 Theright click context menu shown for an Im object S PLUS Functions Called by Statistics Dialogs 344 This context menu implementation allows for application of a diverse range of functions to specific objects by right clicking on them maximizing the flexibility of the object oriented environment of S PLUS To create context menus or to modify existing ones by adding more options refer to the chapter on Programming the User Interface in the S PLUS Programmer s Guide The dialogs described in the following chapters correspond to calls to individual S PLUS functions Most of them have a corresponding on line help file which can be consulted for better understanding of ther structure and valid inputs for field entries Most of these functions act as wrappers to other S PLUS functions which embody a particular type of analysis For example the Linear Regression dialog allows the use
2. 366 Results If a data frameisopm and active then by default this data frameis used in thecrosstabulation If some of its columns are highlighted then by default these columns are used in the crosstabulation If none of its columns are highlighted then by default all of its columns are used in the crosstabulation Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This object is of classcrosstabs This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last Im is the most recent linear regression model fit CONTINGENCY TABLE Print Results Check this to print the results in the designated output window Contiequncy Talha 1 Mode Optoma F Omp Urieed Leme F Aon Ch Sguere Teri how Call Propo ons Decimal teen E F Show bagra Taia Cad eea a a 1 Drop Unused Levels Check this to suppress the display of any unused factor level in the contingency table Show Cell Proportions Check this to display in each cell the count as a proportion of the row total the column total and the grand total Show Margina
3. Names are case sensitive so X and x are different names Print Results KRUSKAL WALLIS RANK SUM TEST Check this to print the results in the designated output window Related S PLus language functions kruskal test menuKruskal 397 CHAPTER 17 COMPARING SAMPLES FRIEDMAN RANK SUM TEST 398 Data Results T his dialog performs a Friedman rank sum test on unreplicated blocked data Data must be in the form of a two way layout with factor variables for groups and for blocks in a data frame See the on line help for friedman test for information on assumptions of the test and the null hypothesis To perform a Friedman rank sum test Choose Statistics gt Compare Samples gt k Samples gt Friedman rank sum from the main menu T he dialog shown below appears Fridman Pose Sur Tait a 12 Daa Fargas Cats Frome See An Paseo sini E E Prim Fiinta Troupin Variabla Biorking Vansbis Data Frame Select a data frame Response Variable Select the column that contains the response variable T his must be numeric Grouping Variable Select the column that indicates group membership for the response above This must be a factor variable Blocking Variable Select the factor column that indicates block membership for each response value T his must be a factor variable Save As Enter thename for the object in which to save the results of the analysis If an object with this name al
4. 534 BUILDING FORMULAS 2 O Overview Linear Regression 536 Transformation 538 Cox Proportional H azards 541 M any of the statistical functions in the previous chapters have the option of building formulas Formulas are required for modeling and defining the response and explanatory variables as well as defining the structure of the model For example if you expect a curvilinear relationship between the response and explanatory variable you can specify the polynomial term in the formula In S PLUS there is a specific structure to the formulas the main feature is the presence of a tilde which separates the listing of the response from the listing of the explanatory variables Using the formula builders described in this chapter you can construct a formula without knowing the details of the formula syntax of the S PLUS language Use the Transformation dialog to manage the transformation of variables when building a formula When the Transformation dialog is active the model dialog which opened the formula dialog and the formula dialog itself are unavailable After performing transformations click OK and the new variable list will be sent back to the Formula dialog which will be active once more Two formula builders are described in the following sections M any of the statistical dialogs use formula builders with a slightly different range of options however the options are present in the two builders described here
5. by contour 313 colors 313 contour lines 313 data grids 311 data range 311 data specifications 311 grid data 311 irregular data 200 levels column 313 levels range 313 number of contours 313 output grids 311 contour plots 3D 330 stacked 330 convergence criterion 422 Convert to Character datamenu 114 Convert to Objects 343 Convert to String button 114 converting data types 114 converting column types 114 converting embedded objects 164 Copy 100 copying cells using the clipboard 100 text in scripts 552 copying data 156 correlation matrix 416 442 460 correlations 368 correlations and covariance matrices 368 count column 350 covariance list 501 Cox proportional hazards 513 creating graphs using menus 180 new script 547 new scripts 547 creating and modify toolbars 568 creating and modifying buttons 569 critical point calculation 481 cross validation score 378 crosstabulation 365 cubic B spline smoother 377 cumulative probabilities 355 current data set 89 curve fitting plots 313 curve type 314 data specifications 314 omit constant 314 plot options 314 predicted values 315 curve type curve fitting plots 314 customizing your workspace 597 Cut 100 607 INDEX D data adding to plots using drag and drop 187 formatting 111 from different data sheets 197 presorting 321 data frame 87 DataFramefidd 342 data frames merging 351 data grids contour plots 311 Datamenu 341 data ran
6. 3 kel g Oo fo e N c 2 ak a c D o 2 o oa DB 9O 99 09 09 09 09 09 09 09 09 09 00 SKID OG 10 GO OH Ol oom NE Per Capita Income Figure9 1 A Trelis graph In S PLUS all graphs can be conditioned using Trellis graphics The data columns used for the plot and for the conditioning variables must be of equal length T he axes specifications and panel display attributes for example fill color are identical for each panel although axis ranges may be allowed to vary T he border and fill attributes for the panels can be specified on the Fill Border page of the Graph Properties dialog 208 CREATING TRELLIS GRAPHS CREATING TRELLIS GRAPHS To create Trellis graphs using the Graph Properties dialog 1 Create a graph 2 Open the Graph Properties dialog by double clicking on the empty space within the graph area Place conditioned plots in separate panels by setting Panel Type to Condition and choosing a conditioning variable from the Column List Click OK To create Trellis graphs using plot buttons 1 Press the Conditioning M ode button on the Standard toolbar and make sure that the desired plot palette is open Notice that the plot buttons now have rectangular strips across the tops of them to denote that you are in conditioning mode On the toolbar set the number of conditioning variables to the desired number Figure 9 2 The Conditioning M ode button and drop d
7. 6 iste ieee eae p Save Dan Frana Properien as chet Help Figure 22 17 A context menu with the item Summary added TheSummary Statistics dialog appears 3 By default Data Frameis set to air in that dialog Click OK and the Summary Statistics are sent to a Report window unless the Command window is open to receive them Instead of the built in Functionlnfo object menuD escribe and its associated built in S PLUS function user defined objects can also be used The procedure for adding a context menu option is identical 4 Applying the Context Menu to a Class which Inherits from data frame 1 Right click on the object catalyst The context option Summary does not appear because the object catalyst has class design which inherits from data frame To confirm this you can check D ata C lass and Inheritance in the Right Pane page of the Object Browser property dialog if this is not already done and view the information in the right pane of the O bject Browser as in figure 22 18 M ake sure that Include D erived Classes is checked in the O bject Browser 593 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE 594 property dialog dap hst dahang Catan daiane BBA harri Oe ane Hera ign deng darne dete kasme des hanes tap beme dmaiheame dstebeame deena danbenc deiehasmo daa bene daiihamia date bore dia ham dsa kame dahana 5 5 5 4 k 5 5 5 Figure22 18 The Object Browser Snowi
8. Context If the cursor is at the beginning in the middle or at the end of aword ina int Script window typing the F1 key will pop up help for this word Specifically Sensitive Help if the word is the name of an S PLUS function help will be shown for this function Using the Right clicking in a Script window will show the following pop up menu T he Right Button Cut C opy Paste C lear options may be enabled or disabled depending on the Menu 555 CHAPTER 21 WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS existence of a selection in the window Undo Cut Copy Paste Clear Select All Show Wala P xiang place Font Help Figure21 8 The Context menu in the Script window Show Dialog 556 If you do not remember the arguments of a function Show Dialog will present you with a dialog box where you can easily fill in the arguments and get on line help Type the name of the function double click on it to select it then right click and select Show Dialog from the pop up menu The dialog for the function will appear Fill itin click OK and your function call isinserted automatically for you in the Script window A similar feature is available to access the dialogs for the object oriented graphics of S PLUS If you need help to generate a line plot from a Script window type the class of graphics that you need for instance LinePlot or Histogram Select the class by double clicking on its name ri
9. D ata Set Specify the name of the data set containing your data Columns Specify the names or column numbers if in a data frame of the data you wish to examine D raw Histogram Choose whether to have histograms shown on the display Spin Window Choose whether to have a spin window shown on the display If you are examining fewer than three variables no spin window will be shown Display options for font and color for Brush and Spin can be found in the O ptions Graph dialog For more information on the use of Brush and Spin see the on line documentation for the function brush A button labeled Quit appears near the middle of the right column of the window just under the bottom of the Spin window Click on the Quit button or choose Close from the Control menu to exit Brush and Spin Inside the Brush window the mouse cursor appears with a box called the brush when it passes inside of any of the scatter plots 1 Move the mouse cursor into one of the scatter plots so the brush covers some data points and click the left mouse button S PLUS highlights all of the points under the brush Simultaneously the points belonging to the same cases are highlighted in all of the scatter plots 2 Now without moving the mouse click the right mouse button the highlighted points downlight back to their original state 3 Highlight additional points with the left mouse button and watch the row of histograms Each histogram
10. Font Page 248 overlap When labels overlap S PLUS first shrinks them to the amount specified to accommodate them If labels still overlap refer to Overlap Handling in this section Horizontal and Vertical Justification Specify the justification of labels relative to the tick marks H orizontal positioning includes Auto Left Right Center and Corner vertical positioning includes Auto Up Down Center and Corner For example if Left is selected for the horizontal position for the x axis all x axis labels will be written to the left of the corresponding tick mark Corner is most useful for angled labels and will align the closest corner of the label with the tick mark Horizontal and Vertical Offset Specify the width of the gap between tick labels and tick marks Auto will position the tick labels relative to the axis position and tick mark length T he offset value is measured in relative inches cm and the label position is measured from the tick mark For example if you specify a vertical offset of 0 25 for the x axis the top of the x axis labels will be positioned 0 25 inches below the bottom of the tick marks Labels can be shifted by a positive or negative amount On the Font page you can specify the font size color and styles for the axis labels Rotation Specify the angle at which to draw your tick labels T he default angle is 0 meaning the tick labels are drawn parallel to the horizontal axis You can specify any
11. He Figure5 5 TheODBC export dialog D etails on the dialog fields follows ODBC Data Source If the database required is not available from the drop down list use the button to the right of it to configure a new data source 138 EXPORTING ODBC FILES Table Name This identifies the name of the table that will be written to in the selected data source If the table already exists you will be asked if you wish to overwrite the existing table or specify a different name Within the Export From group are the following options Data Set Specify the name of an existing data set whose data are be exported Columns and Rows Specify a subset of columns and rows to be exported If no columns or rows are specified or the keyword ALL is specified all columns and rows in the data set will be included in the file 139 CHAPTER 5 IMPORTING AND EXPORTING DATA 140 USING THE OBJECT BROWSER Overview of the Object Browser Overview of the O bject Browser 141 The Right Pane Versus the Left Pane 142 Filtering O bjects and D atabases 143 T he Browser Page Versus the O bject Browser 144 Organization of O bjects 144 The D efault O bject Browser 146 The Examples O bject Browser 146 To Find S PLUS O bjects 146 Shortcut keys 147 Customizing the O bject Browser 148 Customizing the Right Pane D isplay 149 Customizing O bject Browser Pages 151 Using Folders to O rganize Work 153 Editing O bjects and D ata M anipulatio
12. Text to Use C hoose from Specified Text x y z w or O ther Column Choose Specified Text to type text into the Symbol Text box and use it for every symbol on the graph Choose one of the column options to use a column of text as symbols Symbol Text Specify the text to be used as a symbol Column Specify a column of text to use D ata Set Specify the data set containing the column of text Font Choose a font type for the text Bold Italics Underline Specify text formatting options You can add extra dimensions to line and scatter plots by varying symbol Sizes T he relative size of the symbol is determined by the relative value of a third column For example if x is the gas mileage of a car and y is a measure of safety symbol size could vary depending on price The third column can be put in the z or w column on the D ata To Plot page or specified here Vary Size By Choose from None x y z w or Other Column Minimum H eight Specify the minimum size for the symbol PLOT PROPERTIES Smooth Sort Maximum H eight Specify the maximum size for the symbol Column Choosea column from thelist D ata Set C hoose a data set from thelist You can also show an extra dimension of data by varying symbol colors As with varying size the color of the symbols is determined by the relative value of a third column If the third column is a factor the colors will vary according to the colors on the symbols page of the O ptions Graph S
13. 431 CHAPTER 18 FITTING STATISTICAL MODELS 432 Plot Page Plots Options Residuals vs Fit Check this to display a plot of the residuals versus the fitted values Residuals vs Index Check this to display a plot of the standardized residuals versus the index of the observations Residuals Normal QQ Check this to display aN ormal quantile quantile plot of the LT S residuals Residuals vs Robust Distance Check this to display a plot of the LTS residuals versus Robust D istances of x FOWS Number of Extreme Points to Identify Enter the number of extreme points that will be identified on the Residuals vs Fit and Residuals Normal QQ The row names from the data frame specified on the model page will be used to identify the points S PLus language functions related to Least Trimmed Squares Robust Regression Itsreg itsreg formula Itsreg default plot Its summary Its Other related S PLus language functions Im Imsreg rreg LOCAL REGRESSION LOCAL REGRESSION T his dialog fits a locally weighted regression to a response surface To perform local regression Choose Statistics gt Regression gt Local from the main menu The dialog shown below appears Local R egriiiikii i Model Opiore Besubs Pins Predict Data Model Emig Fearne Famiy gum Wieaghts r aiat Subasi Free veh F Ajd Dusit Tem iniLocel Fitmg E Ormia h aring vakaa Drap Dai Temi a F Homair Herens Poschctons Con
14. Data Specs Page In the Import Data dialog the D ata Specs page has an upper section that allows you to navigate using the standard Explorer interface to a particular directory and file to be imported Date Saci Options Filg Loak ji E Hama ml g F an Fik gama Open Fibs ol yes Cancel import To Help Cente Frari Tuga Stan Got i xl Erie Oak I camia Figure5 1 Thelmport D ata dialog D ata Specs page above O ptions page below THE DATA IMPORT DIALOGS Diem Secs Opions ike Inport Opine ASCH ASCH Column Fomes Ciphionel ass Doema Hamax Shari Column BO O O E Erd gome EnD Fomai Ering Erd Paver END Caprera ul File gama Open Files ol yee Tae S01 fa Faken pm bet a z Cancel import To ee OE Tega Sar Got I z Eres y Lf carina Figure5 1 Thelmport D ata dialog D ata Specs page above O ptions page below Options Page In the Import D ata dialog the O ptions page has the following options Name Row If the file you are importing contains names for the columns of data S PLUS can use these names as column names In the N ame Row field specify which row number in the file being imported contains the column names If you do not specify a named row S PLUS strives to locate column names from the first row of the file Specify Row 0 to have S PLUS not search for aname row In a delimited ASCII file the name row must come before the first data rows to be read in
15. Editing Variable N ote that the imported SAS file has variable names all in uppercase and at Names most eight characters long In S PLUS variable names are highly flexible so let us modify the variable names before creating a graph 1 In theright pane of the O bject Browser double click on the column name for RADIATIO and changeit to RADIATION you can use lower case or mixed upper and lower if you prefer 2 Doubleclick on the column name for column 3 and change TEM PERAT to TEM PERAT URE Inserting a New Column Insert Columns Of x Name s Wind Fill Expression JlogWINDD SSSS Start Column END Count i gt Target Eserim Column Type H Cancel Apply i current Help Figure3 13 Thelnsert Columns dialog To create a new column right click inside the data frame on one of the cells and choose Insert Column from the pop up menu or choose Insert C olumn from the main menu and fill out the dialog For instance to create a new column LW ind which is the log of WIND and to insert it at the end of Exenvirn 1 In theN ames field typeL Wi nd 2 In the Fill Expression field type og WIND 3 Leave the other fields at their defaults and click OK to create the new column 66 IN DEPTH TOUR Creating a 2D Graph In S PLUS there are several methods for creating graphs You can select data in a Data window or in the right pane of the O bject B
16. Specify the desired font for your comments as well as color size and styles Rotation Specify the angle at which to draw your comments T he default angle is 0 meaning the comments are drawn parallel to the horizontal axis You can specify any angle between 0 and 360 degrees and labels are angled counter clockwise accordingly PLOT PROPERTIES Contour Level Plots 2D Data to Plot Gridded and Irregular Data Gridding Contour and level plots can use gridded and irregular data Gridded data can be specified in rectangular matrix form or stacked where the Z data is given in one long column and X Y and Z values are given in ascending order Irregular data is specified in a rectangular matrix form You will get better results if the data are distributed fairly uniformly in X and Y and do not contain sharp spikes or drops in Z If only X and Y are specified and they are of equal lengths a2D histogram will be computed for the Z values See section Preparing Data for Graphing page 194 for details on specifying gridded and irregular data X and Y Data Grids Specify the actual number of X and Y grids If you choose Auto S PLUS determines the number of grids from the number of rows and columns in Z or by the number of X and Y values If the X and Y data specifications are left blank and the Z data are in one long column the number of X and Y grids corresponding to the Z data must be specified here
17. eh current Help Data Data Frame Select or enter the name of a data frame gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame x Variable Select the x variable y Variable Select the y variable 370 CORRELATIONS AND COVARIANCES Options Results Smoothing Span Parameter Select a number between 0 and 1 that will be used to control the amount of smoothing Smaller values result in less smoothing Very small values close to 0 are not recommended Degree of Locally fitted Polynomial Select the overall degree of the locally fitted polynomial a 1 is locally linear fitting and 2 is locally quadratic fitting Family Select either symmetric or gaussian The symmetric option combines local fitting with a robustness feature that guards against distortion by outliers T he gaussian option strictly employs local fitting methods No of Values Evaluated Enter the number of points at which to evaluate the loess curve Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names This obje
18. gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame Weights 456 GENERALIZED LINEAR MODELS Formula Model Enter the column that specifies weights to be applied to all observations used in the linear regression To weigh all rows equally leave this blank Subset Rows with Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the model If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Formula Enter a formula specifying the desired model The formula specifies which regression model is to be fit In its simplest form a formula consists of the response variable a tilde and a list of predictor variab
19. Background Color Specify the background color for the Commands window Echo If checked each complete expression will be echoed before it is evaluated Length Specify the length in lines of an output page Main Prompt T he string to be printed to prompt for an expression default is gt Continue Prompt T he string to be printed to prompt for the continuation of an expression default is To specify Undo and History options choose Undo amp History from the Options menu The Undo amp History dialog will appear of Undos Specify the maximum number of undos to be saved in the undo queue this does not apply to undoing changes to data objects nor to text edits in scripts The higher the number of undos the more memory is used History Entries Specify the maximum number of entries to be saved in the history log The higher the number of entries the more memory is used CHANGING DEFAULTS AND SETTINGS Specifying Text Output Window Settings Specifying Graph Options Options Page To specify text output settings select O ptions Text O utput from the menu bar The O utput window preference dialog will appear Output Window Preference C hoose whether to have S PLUS use its default settings when redirecting all output or use the Commands window the Report window the Script program window or the Script output window To specify graph options select O ptions Graph O ptions from the menu bar TheGra
20. Changing the Plot Type Plot Types 295 Changing the Plot Type 295 Plot Properties 299 Common Plot Properties 300 Area Charts 303 Bar Charts Vertical and H orizontal 2D 304 Box Plots 307 Comment Plots 309 Contour Level Plots 2D 311 Curve Fitting Regression 313 Error Bar Plots 316 H igh Low C lose Plots 317 Histograms and D ensity Plots 318 Line amp Scatter Plots 2D 319 Pie Charts 325 Polar Plots 326 Vector Plots 327 Scatterplot M atrix 328 Contour Plots 3D 330 Line amp Scatter Plots 3D 330 Surface Plots 332 This chapter describes the detailed changes that can be made to modify the appearance of your plots See section S PLus graphs page 174 for a discussion of the plot palettes In S PLUS you can switch between different plot types using the plot palettes or the Change Plot Type option on the Format menu If you choose a plot type that is incompatible with your data selection you will need to change 295 CHAPTER 12 EDITING PLOT PROPERTIES 296 your data specifications If you change the plot type to one with a different axis system you will be prompted before the graph is converted Under these circumstances you will lose axes specifications and other contents of the original graph To change the plot type using a plot palette 1 2 Select the plot you wish to change Click the 2D or 3D Plots button A palette of available 2D or 3D plot types appears Click the desired plot button
21. T his plot produces side by side box plots from a number of vectors T he box plots can be made to display the variability of the median and can have variable widths to represent differences in sample sizes This plot uses the function boxpl ot for the underlying computations Single box plot Specify the name or column number of your Y column no X column is specified A description of the data values in the Y column are displayed in a single box plot The X position defaults to 1 but can be changed Grouped box plot Specify the names or column numbers of your X and Y columns The X column is a numeric column that assigns each Y value to a group T he data in the X column also determines the placement of the boxes along the x axis Data in the X column should be integers any non integer values are truncated The Y column contains the data values to be used in computations 307 CHAPTER 12 EDITING PLOT PROPERTIES 308 Specifying data using a long X and long Y column In the long form the X column is a numeric column that assigns each Y value to a group T he data in the X column also determines the placement of the boxes along the x axis Data in the X column should be integers any non integer values are truncated The Y column contains the data values to be used in the computations In this example three boxes would be plotted and placed one interval apart on the x axis Table 12 2 Long form data X Y I 0 25
22. Table 21 3 Optionsin the Display H isory dialog Option Purpose Script Name Start with Entry End with Entry Selected O bject O nly Reverse O rder Specify a name for the script that will contain the H istory log This is optional the default script name is HISTORY Specify the starting and ending entry numbers to be displayed in the History log This lets you control the number of History entries and which entries get placed in the H istory log Choose to have the script contain History entries for the selected object only This is useful when you want to focus on the commands for a specific object For example if you select a symbol the script will contain all the entries related to creating and modifying the symbol Choose to display the H istory entries in the reverse order in which they were generated You will see the command executed most recently at the top of the script To execute recorded commands in the H istory log 1 Usethe mouse to highlight H istory entries you wish to execute 2 Click the Run button on the script toolbar or choose Run from the Script menu You can also cut paste and save these commands to another script file To clear the History log You should clear the History log before you start recording steps T his will save editing time later and make it clearer what commands were generated by the actions you made in menus and dialogs w From the Window menu ch
23. Variable Selection Compare Levels Of Select the term in the model to which comparisons will be made T his list will be empty until a selection has been made in M ode O bject Method Options Compare To Level Select the factor level to which all other levels will be compared T his field is available only when Comparisons is set to mec Comparisons Select the type mca e mcc none Method of comparisons to be made among the adjusted means all pairwise differences all pairwise differences between all adjusted means and the adjusted means for the factor level specified in Compare To Level if the adjusted means themselves are of interest without further differencing Select the method to use for critical point calculation e best best fast Bonferroni Dunnett Fisher Isd Scheffe Sidak Simulation for the smallest critical point choosing from all valid methods for the smallest critical point choosing from all valid methods except Simulation for the Bonferroni method for D unnett s method for Fisher s unprotected least significant difference method for Scheffe s method for Sidak s method for an approximate critical point using simulation based methods 481 CHAPTER 18 FITTING STATISTICAL MODELS Error Specification 482 Results Tukey for Tukey s studentized range method See the on line help for mul t i comp or the chapter M ultiple Comparisons in the Guide t
24. acceptance and transplant event Variablesstart and stop mark the time interval and the variableevent isthe event indicator for the stop time 543 CHAPTER 20 BUILDING FORMULAS 544 1 GE See ae US Choose Statistics Survival C ox Proportional H azards from the main menu In the D ata Frame field enter or select heart Select counting from the Type of Censoring drop down list Click the Create Formula button Create the Survival Response by selecting start from the Choose Variable s list and click the Time 1 button Select stop from the Choose Variable s list and click the Time 2 button Select event from the Choose Variable s list and click the Censoring Indicator button Click the Add Response button Define explanatory terms by holding down the CTRL key while selecting age and transplant from the Choose Variable s list T hen click on the M ainHnteract button from the Add Term s group Select year and surgery from the Choose Variable s list Click the M ain Effects button from the Add Term s group A survival formula using two variables which mark the time interval is now created Click the OK button to accept the formula and return to the C ox Regression dialog Similar Formula Builders T he following statistical models use a similar Formula Builder as that for C ox Proportional H azards Parametric Survival Nonparametric Survival WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS Overview Overvie
25. or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only thefirst 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Treat Missing with 365 CHAPTER 16 SUMMARIZING AND EXPLORING DATA Formula Select an option to handle missing data e nafail If any missing values are encountered the procedure fails with an error naomit Cases with N Asin any of the variables to be crosstabu lated are omitted from the crosstabulation nainclude The level NA is added to each factor before the crosstabulation is performed It is also possible to enter a user defined function here If there are any miss ing values in the data to be crosstabulated the data will be put into a data frame and passed to this function Factors Select one or more columns to be used in the crosstabulation Counts Variable Use this only if each row represents multiple observations Select the column which specifies the number of replications corresponding to each row Formula When Factor or Counts Variable is updated this field is automatically updated By default all variables are used in the crosstabulation and each row is taken as one observation T his is indicated by in Formula T his field can also be edited manually gt Tip
26. 320 symbols styles 603 system requirements 18 T ttet 388 tabular summary 349 technical support 21 third quantile value 363 time series 526 estimate type 526 forecasts 532 time series data 98 tool tips disabling 598 599 Toolbar property dialog 572 ToolbarButton dialog Image page 571 T oolbarButton property dialog 569 toolbars color 599 hiding and unhiding 572 toolbars and palettes customizing 568 training courses 20 transformation 538 tree models 63 473 pruning or shrinking 477 tree regression 473 Trellis graphics 59 Trellis graphs 210 examples 216 formating panels 212 M ultipanel page 212 panel strips 214 trickling input 565 trouble shooting plot redraw is sow 604 two sample K olmogorov Smirnov test 392 two sample t test 388 two sample W ilcoxon test 390 U undo actions 108 selective 108 Undo button 108 Undo options 600 undoing 108 user defined colors 603 V variable spans 373 Variables field 342 variance estimate 363 615 INDEX vector 88 vector plots 327 angle units 328 angle magnitude 328 begin end 327 data specifications 327 data type 327 magnitude multiplier 328 plot options 327 vector position 328 vertical error bars 316 viewing in draft mode 182 W weight line 301 symbol 302 616 weighted averaging 335 Wilcoxon test 382 390 Win32s 16 Windows 3 1 16 W indows for W orkgroups 3 11 16 workbox defined 254 X x axis tyle 511 519 Y Yates
27. Bold Italics Underline Specify text formatting options You can add extra dimensions to line and scatter plots by varying symbol sizes and colors Vary Size By Choose from None x y z w or Other Column Minimum H eight Specify the minimum size for the symbol Maximum H eight Specify the maximum size for the symbol Column Choosea column from thelist D ata Set C hoose a data set from the list You can also show an extra dimension of data by varying symbol colors As with varying size the color of the symbols is determined by the relative value of a third column The colors are specified as a range or as a list of special colors Vary Color By Choose from None x y z w or Other Column Colors To Use Choose from Range or Special Minimum Color Minimum color in the shading range Maximum Color M aximum color in the shading range Column Choosea column from thelist D ata Set C hoose a data set from the list Special C olors Specify custom colors D raw Plane C hoose whether to display a regression plane X and Y Grids Specify the number of X and Y grid lines T he density for the regression plane grid depend on the values specified here The higher the values the denser the grid 331 CHAPTER 12 EDITING PLOT PROPERTIES Regression Plots 3D Surface Plots Data to Plot Gridded and Irregular Data Lines 332 Plane Attributes Specify the line color style and weight for the lines used to draw the regre
28. D ay Series Year Series Long Date Short D ate Long Date Time Short Date Time Time Elapsed H M S M S 246 Specify that the labels be in financial format Specify a month name in the Start field and an interval number in the Interval field for example January and 2 Specify a day name in the Start field and an interval number in the Interval field for example Sunday and 2 Specify a year in the Start field and an interval number in the Interval field for example 1981 and 2 Converts column of numbers to day of the week month day year Converts column of numbers to mm dd yy format Converts column of numbers to day of the week month day year hh mm ss Converts column of numbers to mm dd yy hh mm ss format Converts column of numbers to hh mm ss format Choose to specify data in either hh mm ss or mm ss format Precision for Decimal Scientific or Mixed only Specify the number of decimal places to display after the decimal Start Value for M onth D ay Year Series only Specify a starting value for the series Series Increment for M onth D ay Year Series only Specify an increment value for the series Column For Column labels only Specify the name of a column containing the labels The first row in the column is used to mark the first major tick mark the second row the second tick mark and so on You can leave any number of rows blank For example if you want labeling to begin
29. D esigned to provide a unified standard way to exchange data between databases OD BC has become widely supported Each application typically has an O D BC driver that allows the application to accept or distribute data via the ODBC interface S PLUS supports OD BC versions 2 0 and 3 0 You may already have the ODBC Data Source Administrator hereafter referred to as the Administrator installed on your personal computer You can verify this by opening the Program Manager Windows 3 x and Windows NT 3 51 For more recent versions of Windows select Settings then Control Panel from the Start M enu O pen the Control Panel and see if it contains an ODBC applet If so you can skip the rest of this section unless you want to upgrade from version 2 0 to version 3 0 of the Administrator you may still need to install OD BC drivers however The S PLUS CD includes everything necessary to install version 3 0 of the ODBC Administrator on all current Microsoft Windows platforms M athSoft does not distribute any drivers so you will need to provide your own drivers to access the data source you wish to connect to Contact the vendor of your database to learn what drivers they provide or contact other third party vendors who sell ODBC drivers For instance for M icrosoft O ffice 97 select the D atabase D rivers option in the D ata Access component during set up M icrosoft Access users may also need to install the D ataAccess Pack located in the Valupac
30. D own arrow move down to next tree item move down to next list item Left arrow collapse selected tree node Right arrow expand selected tree node Home go to first element in tree go to the first element of the list End go to the last dement in thetree go to the last element of the list Tab make the right pane active make the left pane active Shift Tab activate the next tab activate the next tab Return same as double click action of same as double click action of selected object selected object Ctrl C copy selected object copy selected object s Ctrl V paste selected object paste selected object s Delete delete selected object delete selected object s 147 CHAPTER 6 USING THE OBJECT BROWSER CUSTOMIZING THE OBJECT BROWSER 148 The Object Browser may be customized to contain multiple Browser Pages each filtering on different S PLUS data bases and object classes M oreover the right pane of the O bject Browser can be modified to display various object characteristics O nce a desired configuration is found it can be saved to afile T he file extension used for the O bject Browser configuration files is SBF To create an empty O bject Browser use the File N ew menu item and select Object Browser from the New dialog An empty Object Browser will be created To modify an existing O bject Browser select the Format O bject Browser menu item or right click in the right pane of the O bject Browser and choose Browser from the context
31. D rag left to increase column width drag right to decrease column width Access context sensitive online help Displayed when M ove or Size is selected from the C ontrol menu allows the window to be moved or resized Change the size of the window vertically or horizontally when positioned on a window border Change the size of two sides of the window when positioned on the corner of a window border Indicates that a command is being processed you should wait for a different mouse pointer before going on to other tasks T he current command is being processed and cannot be interrupted A drag isin process and the object cannot be dropped successfully in the current position You must move to another position to drop the object Throughout this User s Guide the following conventions are used to reference keys Key names appear in all uppercase letters For example the USING MENUS DIALOG BOXES AND TOOLBARS Using Windows Shift key will appear as SH IFT W hen more than one key must be pressed simultaneously the two key names will appear with a plus between them For example the key combination of sHIFT and F1 will appear as SHIFT 1 Theup down left and right direction keys represented on the keyboard by arrows are useful for moving objects around the page T hey will be referred to as the up direction key the pown direction key theLert direction key and the RIGHT direction key In S PLUS you can o
32. Hint Refer to the Programmer s Guide for details of the syntax of formulas should you need them and to the Guide to Statistics for many examples of formulas used in a wide range of statistical models 535 CHAPTER 20 BUILDING FORMULAS LINEAR REGRESSION Use the Create Formula button from the model page of the Linear Regression dialog to open up the Formula page Yenshie Spec Tem Chin ae vasablajsi ae Tarh Cahir T CEE i r m UE Tna aj Fomai Trenefo eto Add Adi Tamy Aspa Mein Eeti DV Remir hiereg e E Cusine fa 2 Cubic fe Parteseas Powrrals Fui Aa oy lea ray Figure20 1 Formula page for the Linear Regression dialog Tip A data frame is required to open the formula page When the formula pageisup the D ata Frame and the Formula group in the moda page are disabled until OK or Cancel is clicked If a formula is composed and OK or Apply is clicked the formula issent back to the moda page Variable Choose Variables Variables of the D ata Frame selected in the model page are displayed in the multiple selection box Choose Variable s W hen it is appropriate you may press on the CTRL key to select more than one variable After selecting variables assign the selection as response or predictors to complete the formula If you want to do transformations on some variables click on Transformation 536 LINEAR REGRESSION Add Special Term Remove Cl
33. Names are case sensitive so X and x are different names It will be saved as a list consisting of the fitted smoothing spline evaluated at the supplied data some fitting criteria and constants and a structure that contains the essential information for computing the spline and its derivatives for any values of x Plot Smooth Check this to display a scatterplot with the Spline smooth Related S PLus language functions smooth spline predict smooth spline print smooth spline COMPARING SAMPLES One Sample OneSample t Test 380 O ne Sample W ilcoxon Test 382 O ne Sample K olmogorov Smirnov Test 384 O ne Sample C hi Square G oodness of Fit Test 386 Two Samples Two Sample t Test 388 Two Sample W ilcoxon Test 390 Two Sample K olmogorov Smirnov Test 392 k Samples One Way Analysis of Variance 394 Kruskal Wallis Rank Sum Test 396 Friedman Rank Sum Test 398 Counts and Proportions Exact Binomial Test 400 Proportions Test 402 Fisher s Exact Test 404 M cN emar s C hi Square Test 406 M antel H aenszel C hi Square Test 408 Pearson s C hi Square Test 410 Introduction This chapter describes some dialogs to test hypotheses involving one sample problems two sample problems k sample problems and distributional goodness of fit These include dialogs for parametric and non parametric tests for continuous as well as discrete data Refer to chapter 14 Using the Statistics Menus and Dialogs for an explanation of features common to mo
34. Predictions Check this to save the predictions in the data frame specified in Save In Standard Errors Check this to store the pointwise standard errors for the predictions in the object specified in Save In 471 CHAPTER 18 FITTING STATISTICAL MODELS Options Prediction Type Select the type of prediction to be saved e link predictions are on the additive predictor link scale response predictions are on the response scale using the inverse link function terms a matrix of predictions is produced one for each term in the model S PLus language functions related to Generalized Additive Models gam plot gam plot glm predict gam summary gam Other related S PLUs language functions gim loess ace avas 472 TREE MODELS TREE MODELS This dialog can be used to fit classification and regression trees to data in a data frame See also chapter 10 Classification and Regression Trees in the Guide to Statistics See also on line help for thet ree function To perform tree regression Choose Statistics gt Tree Models from the main menu The dialog shown below appears Tine Amgmasion E Model Reme Pict Preneysitnns Pradig Dem Fitting Opnar Cuis Frame bin Ho o tes Defme Sphi Wieoghts P ibas Fire wh oOo bin Node Sirs fia P Omi Foes wih Misaing vekas Min Hode Derisnceft M Ima Mida Ol bat Seve An faa wae Fomuls Faridi Crosio Fore Model page Data Data Frame Select a dat
35. Scale Factor for Y Axis Tick Label Enter a number used to multiply the labels on the y axis For example a value of 100 yields a percent scale 511 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS 512 X Axis Label Enter a label for the x axis Y Axis Label Enter alabel for the y axis Related S PLus language functions for Nonparametric Survival survfit print survfit plot survfit points survfit lines survfit summary survfit survfit km Other related S PLus language functions coxph Surv strata COX PROPORTIONAL HAZARDS COX PROPORTIONAL HAZARDS T his dialog fits a C ox Proportional H azards regression model to survival data Timedependent variables timedependent strata multiple events per subject and other extensions are incorporated using the counting process formulation of Andersen and Gill See also the chapter 22 The Cox Proportional H azards M odel in the Guide to Statistics To perform Cox regression modeling Choose Statistics gt Survival gt Cox Proportional Hazards from the main menu The dialog shown below appears Cox Brn netinnal Hixenede S 4 es e k mE E Moda Options Freud Predict Survived Cotes Cista Model Deis Fremo Type of Careming fighi mag his Sebsetowswh Saa Modd Object Sav Aa fasi reps F mafi wan Merging aheg Fornis Fomu FE Create Forms Model Page Data Data Frame 513 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND
36. Simm ji bainiie Dyar Wich io Colapps CV Pradiimis M Sorda Errors Predict Page New Data Save Enter the name of a data frame containing the specification of future covariate history for the patients in question T he columns of the data frame must have the same names as those used in the model formula Save In Enter the name of an S PLUS data frame in which a part such as predictions and standard errors of the analysis is saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame T his allows you to keep predicted values from a model with the original data or to keep the predictions from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Predictions Check this to save predictions 517 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS Standard Errors Check this to save standard errors Type Select the type of model predictions Ip linear preditors risk expected or terms Variable Over Which to Collapse Enter the name of a grouping variable over whic
37. This applies to Property objects having Type Page or Group Tip A Property object may only be called once by a given FunctionInfo object Copy From Enter the name of a Property object to be used as a template The current Property object will differ from the template only where specified in the property dialog See the section Internal Resources below for a list of internal Property objects that can be used in dialogs via Copy From Is Required Check here to require the Property object to have a value when OK or Apply is clicked in the dialog Use Quotes Check hereto force quotes to be placed around the value of the Property object when the value is passed to the S PLUS function No Quotes Check here to prohibit quotes from being placed around the value of the Property object when the value is passed to the S PLUS function T his option is ignored when Is List described below is not checked Is List Check here to cause a multiple selection in a drop down list to be passed as an S PLUS list object to the S PLUS function No Function Arg Check here if the value of this Property object is not passed as the argument to the S PLUS function The Property object must still be referenced by the Function nfo object Disable Check here to cause the Property object to be disabled when the dialog starts up Help String Enter the text of a tool tip for this Property object IsReadOnly Check here if the corresponding contr
38. Variable Sid Dasaim Berie Emaii E Friii Rawets Select the column for which the one sample Kolmogorov Smirnov test will be applied Hypotheses Alternative Hypoth Select the alternative hypothesis For example to perform a one sided test 384 ONE SAMPLE KOLMOGOROV SMIRNOV TEST Parameters of the Selected Distribution Results against the alternative hypothesis that the hypothesized distribution is greater than the true distribution of the variable select greater from the list box Hypothesized Dist Select the hypothesized distribution for the variable W hen a hypothesized distribution is chosen the parameters associated with the distribution are enabled For example if the Cauchy distribution is chosen the Location and Scale are enabled Enter the location and scale parameters of the C auchy distribution Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last lm is the most recent linear regression model fit Print Results Check this to
39. item s context menu in the O bject Browser After that the dialog does not appear and the function executes using the current dialog defaults function associated to the menu item and dialog refers to the dialog associated to that function M ove to the I mage page of the dialog Image FileName Enter the path and filename of a bitmap file whose image will be displayed on the toolbar button U se the Browse button if desired to identify the file To modify a ToolbarButton object use either the ToolbarButton property dialog described above or the context menu described below 571 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE Modifying Toolbars EA Fe ea I me Using the Context Menu Insert Button Select this to insert a new toolbar button next to the current one Hide Select this to hide the toolbar button D elete Select this to delete the toolbar button Edit Image Select this to open the bitmap file using the operating systems default bitmap editor which contains the icon image of the toolbar button Button Select this to open the Button page of the property dialog for the toolbar button Command Select this to open the Command page of the property dialog for the toolbar button Image Select this to open the Image page of the property dialog for the toolbar button Save ToolbarButton Object as default Select this to save a copy of the ToolbarButton object as the default ToolbarButton object Hel
40. line color of the plot to green in the third panel and to bluein the rest of the panels To edit plots in individual panels 1 In theleft pane of the O bject Browser select GraphSheet 2 Expand the branches until you see the plot panels appear as individual objects 3 Doubleclick on the panel you wish to edit to open its property dialog M ake the desired changes and click OK FORMATTING 2D AXES FORMATTING 2D AXES Adding an Additional Axis You can add additional axes to your 2D graph You can drag and drop them from the 2D plot palette or you can use the Axis option on the Insert menu To add an additional axis Figure 10 3 The extra axis buttons on the2D plot palette 1 or D rag one of the extra axes off the 2D plot palette and drop it inside the graph area An extra axis is added to the selected graph If an axis already exists in that position the new axis is automatically offset slightly from the original axis Select the graph From the Insert menu choose Axis From the Axis submenu choose U pper X Axis Lower X Axis Left Y axis or Right Y axis The Axis dialog appears Specify the position and formatting attributes for the new axis and choose O K T he extra axis is added to the selected graph To add an axis with a frame 1 or D rag one of the axes with frames off the 2D plot palette and drop it inside the graph area An axis and frame are added T he frame is not atrue axis it is
41. nearest to each grid point and then finds where the plane through these points intersects the vertical line that goes through the grid point that is the line perpendicular to the x y plane at the grid point This intersection is 311 CHAPTER 12 EDITING PLOT PROPERTIES Labels 312 used as the estimate of the height of the surface at this point Weighted Average Used only for the internal algorithm S PLUS will calculated a weighted average of all points that lie within each grid cell before interpolating the grid height If your irregular data has many irregularities using the weighted average method may smooth them out within each grid cell As a result the weighted average method slows down plotting but produces a smoother contour If the average number of data points within each cell is less than three there is not likely to be any advantage to using weighted averaging Extrapolate U sed only for the bivariate algorithm If checked data points will be extrapolated outside of the convex hull determined by the data points filling the entire computed grid If not checked only the area of the plot corresponding to the convex hull of the original data will be drawn The internal algorithm always extrapolates Pts Partial Der Used only for the bivariate algorithm Percentage of additional points to be used in computing partial derivatives at each data point If set to zero partial derivatives are not used The implied number
42. specify Weight 1 and Tick Position In 4 Click OK Then repeat steps 1 to 4 for the x axis 5 Change the x axis title by clicking once to select the axis label and once more to get an in place edit box 6 In the text edit box type O zone Concentration Click outside the box to make the change Change the Y axis title to Solar Radiation 1 Doubleclick on a point in the plot to display the Line Scatter Plot dialog 2 On the Line page specify the Line Color as Lt Blue and the Weight as 2 On the Symbol page specify Diamond Solid as the Symbol Style and Red as the Symbol Color Click OK IN DEPTH TOUR Legends 1 To add a legend to our graph click on the Auto Legend button located on the Graph sheet toolbar T he legend will appear Figure 3 15 TheLine Color button circled on the Graph toolbar changes text as well as line colors the Auto Legend button is marked with an dli ps Titles 1 Now we can insert a main title at the top of our graph From the Insert menu choose Titles and then M ain 2 Type The Relationship Between Radiation and Ozone into the text box that appears Click outside the title to close the text box 3 Select your M ain Title and use the toolbar to set a font size of 20 Use the Line Color button to change the text color to M agenta Minimize the Graph sheet and D ata window The Reblior hip Betaeen Radalion and Ofer niger Figure 3 16 After making all
43. specify the panels in which you would like your plane to appear or All to haveit drawn in all panels Plane Labels Choose whether to have labels along the plane T he numbers along the corresponding axes are used for the labels Hide Choose whether to hide or display the axes In the 3D Plane dialog the Grids page has the following options Subdivisions H orizontal and vertical grid lines can be drawn at every major tick and at intervals between the ticks Specify the number of subdivisions between tick marks to use for drawing the grid For example if you specify a value of 1 there will be an extra grid line drawn between each tick mark FORMATTING 3D PLANES Fill Border Page You can specify the line style color and weight for the grid lines in each direction Specify none for the line style to have no grid drawn in that direction You can specify standard fill attributes fill color fill pattern and pattern color and standard line attributes style color and weight for the 3D planes 253 CHAPTER 10 FORMATTING A GRAPH ROTATING A 3D GRAPH You can change the angle at which you are viewing your 3D graph When specifying the rotation in the dialog you are specifying the position of an observer of the graph To rotate the 3D graph you must rotate the workbox which isa3D box surrounding the graph To rotate a 3D graph interactively 1 Click the workbox to select it Horizontal rotation knobs round
44. 1 0 45 1 0 98 2 0 32 2 0 10 3 0 05 3 0 78 3 0 34 3 0 29 Specifying multiple Y data columns If your grouped box plot are arranged in multiple y data columns you can specify the grouping column levels for X and multiple Y columns for the Y data specification for example Samplel Sample5 See section Preparing D ata for Graphing page 194 for an example of multiple column data Specifying stacked Y data If your grouped box plot data are stacked the number of rows in column Y must be evenly divisible by the number of rows in column X in order for the data to be arranged in groups The number of rows in X determines the number of groups boxes See section Preparing D ata for Graphing page 194 for an example of short form stacked data PLOT PROPERTIES Box Specs Other Specs Comment Plots Data to Plot For box plot options you can specify standard line attributes style color and weight as well as fill attributes color pattern and pattern color for the box For the median line standard line attributes and symbol attributes style color weight and line weight apply For other options you can specify standard fill attributes color pattern and pattern color for the confidence bounds For the whiskers standard line attributes are available For whisker caps and outliers standard line attributes style color and weight and symbol attributes style color weight and line weight apply Additional options ar
45. 3 to 12 for the number of tick intervals For histograms and box plots ticks will be drawn at an interval equal to the smallest data interval see D ata below For bar charts ticks are placed at equal intervals with the x data used as labels see Categorical below If you choose Data S PLUS calculates the smallest interval between data points and uses this interval for tick placement For example if your X data is 1 0 1 25 1 75 2 0 and 3 0 the width of the tick interval will be set at the smallest interval between data points 0 25 D ata will produce atick mark at each data point when your data is regularly spaced If you choose Categorical ticks are placed at equal intervals along the axis and the column of data specified in the D ata to Plot page including string data is used as tick labels Minor Tick Placement Interval W hen specifying minor tick marks you can specify either the number of tick mark intervals or an interval size in axes units Specify the number of minor tick intervals between major tick marks for each axis A value of 2 will place one minor tick between each major tick A value of 1 will remove all minor ticks Minor tick specifications are only available for 2D and Polar Axes M ajor M inor Interval Type Choose N umbers to specify the number of tick mark intervals Choose Size to specify the interval size in axes units Choose Auto to use the default interval type Choose Column to use a specified FOR
46. 360 degrees measured from the x axisin the XY plane A value of 0 gives a viewpoint with the x axis pointing toward you Distance Specify the distance from the center of the workbox to the viewpoint observer in workbox units The number must be greater than zero and should generally not be too small to avoid distortion As the distance increases the size of the graph does not change only the perspective changes For example to position the viewpoint directly behind the graph and centered from top to bottom you would specify 120 degrees for the angle to the z axis 45 degrees for the angle from the x axis and 10 for the distance to the center assuming the workbox x specification is 1 0 Workbox Shape Specify the relative dimensions of the workbox by specifying the X Y and Z size ratios The workbox whether drawn or not determines the relative sizes of the x y and z dimensions of the graph O nly relative values are important For example specifying 1 0 1 0 1 0 results in the x y and z dimensions being the same length the workbox is a cube Specifying 2 0 2 0 2 0 has the same result The workbox shape cannot be changed interactively with the mouse Workbox Attributes and Fill Select the style color and weight for the frame around the workbox and the fill inside the workbox 255 CHAPTER 10 FORMATTING A GRAPH DISPLAYING 3D MULTIPANEL GRAPHS Displaying 3D Rotation Graphs Use multiple panels to display your
47. Axes page 259 for information about polar axes specifications This plot compares the distributions of two data sets by graphing the quantiles of one data set against the quantiles of a second data set T his plot uses the function qq for its computations Specify a single vector to be compared to a theoretical distribution or X and Y vectors to compare against one another Line Symbol You can specify all of the same attributes asa 2D Line Scatter plot Distribution You can specify the standard line attributes style color and weight for the distribution line Function Specify the function for a theoretical distribution The fields following the function field are the input parameters for the available functions The appropriate parameter fields will activate as each function is selected Vary Symbols You can specify all of the same attributes as a 2D Line Scatter plot Smooth Sort You can specify all of the same attributes as a 2D Line Scatter plot Vector plots can be created using the Graph menu on the Standard toolbar For vector plots you need to specify the names or column numbers of your X Y Z and W columns For vector plots you can specify standard line attributes color style and weight and arrowhead attributes Data Type Specify whether your vector data contains the beginning and ending points for each vector or contains the vector position angle and magnitude If you choose Begin End you need to spec
48. Choose M edian of X Y Z or W to arrange panels by median values of the subsamples in each panel of these data columns Choose M ean of X Y Z or W to arrange panels by mean values of the subsamples in each panel of these data columns Number of Columns Rows Pages Control the layout of the panels by specifying the number of columns rows and pages Choose Auto for automatic calculation Vertical H orizontal Margin Specify the vertical and horizontal spacing between the panels in document units Panel Order Choose from Graph O rder or Table O rder Graph O rder begins drawing panels in the bottom left corner of the graph to the right and up Table Order begins drawing panas in the upper left corner and continues right Draw Strip Choose this option to draw a title strip on each panel for each condition Rotate 3D Axes Available for 3D graphs only Select this option to show the plots rotated in 3D spacein different panels See the 3D Workbox page to control the degree of rotation Number of Panels If the data are continuous the number of panels will be determined by the number specified in this field Fraction of Shared Points Create overlapping intervals by specifying the fraction of data points that are shared across two panels Interval Type Choose from Equal Counts Equal Ranges or Columns of Ranges Equal Counts places an equal number of data points in each plot Equal Ranges makes the interval widths all equal Columns of Ran
49. Coady O Ara Maren Eo l Epecified Coli Fined ER Seer r e mey E A F Jisan Daai Data Frames Data Frame 1 Row Matching Select the first data frame from the D ata Frame 1 list box Data Frame 2 Select the second data frame from the D ata Frame 2 list box Match by Select the method to use for matching rows If All Common Cols is selected rows containing identical values on all columns with the same name in both data frames will be merged If Row N ames is selected rows with identical row names in both data frames will be merged If Specified Cols is selected the Columns in D F 1 and Columns in D F 2 fields may be used to specify columns to use for matching Columnsin D F 1 Specify the names of the columns in D ata Frame 1 to use for matching rows Columnsin D F 2 Specify the names of the columns in D ata Frame 2 to use for matching rows T hese column names must be ordered in parallel to the namesin Columnsin D F 1 in order to specify which specified column in D ata Frame 2 to match 351 CHAPTER 15 CREATING AND MANIPULATING DATA Include Non Matched Rows in Suffix for Non Matching Common 352 Cols Results against each specified column in D ata Frame 1 Data Frame 1 Check here to include rows unique to the first data frame Data Frame 2 Check here to include rows unique to the second data set If neither of the above boxes are checked then the resulting data frame will only co
50. Commands are ended by typing RETURN but if the statement is not correctly terminated a continuation prompt is used reminding you that more input is necessary This can be useful when entering long statements Use meaningful names for assignment confusion can reign if single letter names are used For example T is reserved for logical true and t is the transpose function The Up and Down arrows can be used to scroll backward and forward through the list of commands So a typing error can be easily corrected or a new similar command entered without completely retyping it O ther editing commands are described in the first chapter of the Programmer s Guide All commands entered into the Commands window will appear in the H istory log To get help on any S PLUS function just type it in preceded by a question mark For example 2aov will give a help file for the analysis of variants function S PLUS entirely typeq at the prompt THE REPORT WINDOW THE REPORT WINDOW Operations Report window operations The Report window is similar to the Script window T hey both are primarily text windows which can be opened and saved via the File menu and are editable Unlike the Script window the Report window does not deal with programs or scripts The Report window is a placeholder for the text output resulting from any operation in S PLUS H owever the user must select it as the preference for the text output See sec
51. Contidence Interests Model Optora prre Fd perc Fe o wakdiy Chack Cios Point F Emmaa Check Sheri taiann Sine Options Page Contrast Matrix Enter the name of a contrat matrix Each column specifies a linear combination to be estimated under the textbook parameterization of the linear model See the on line help for mul ti comp or the chapter M ultiple Comparisons in the Guide to Statistics for more information Critical Point Enter a value for the critical point used in the confidence intervals bounds Use this if none of the methods are suitable Simulation Size Enter the size of the simulation to use This is available when M ethod is Simulation or Best The default value provides intervals or bounds whose actual family wise error rate is within 10 of the requested rate Validity Check Check this to check the validity of the specified critical point calculation method for the desired comparisons If the validity check fails processing will stop with an error message Estimability Check Check this to check estimability of the desired linear combinations If the estimability condition fails processing will stop with an error message S PLUS language functions related to Multiple Comparisons mul ticomp mul ticomp default 483 CHAPTER 18 FITTING STATISTICAL MODELS multicomp Im print multicomp plot mul ticomp Other related S PLus language functions aov Im 484 COMPARE MODELS COMPARE MOD
52. Creation Modifying Data Objects The O bject Browser gives users access to existing objects so that they can be modified or copied among S PLUS databases N ew objects can also be created from the O bject Browser Empty graph sheets data frames matrices vectors and lists can be created from the O bject Browser T he newly created S PLUS objects are placed in the working database database 1 For the object to appear immediately make sure the Browser Page that initiated the creation is currently filtering on the working database T he context menus for the root node objects labeled GraphSheet data frame matrix vector and list each havea Create menu item that will create an object of the same class The S PLUS data objects are labeled SPx and the GraphSheets are labeled GSx where x is an integer that makes the name unique M oreover all Interface objects can be created from the context menu of any Interface of the same class in the Object Browser The exceptions are the object classes O bjectD efault SearchPath and S PLUS functions Columns of data can be deleted from a data frame or matrix object First select the object in the left pane then select one or more columns contained in the object in the right pane Multiple selection is done by holding down the CTRL key and block selection by holding down the SHIFT key To delete a column right click on one of the columns selected and select the D elete menu item or press the
53. D elete key Columns of data can be copied from one data frame or matrix to another using the data column s context menu or by drag drop In order for the receiving object to accept the column s the number of rows in the receiving object must be a multiple of the length of the column s D ata frames matrices and vectors can be copied into a list object via a copy and paste operation or by drag drop When pasting or dropping an object into alist anew instance of the object will be appended to the end of the list This operation is not valid for sub lists of lists Elements of a list can be deleted in the same manner as that for columns of data frames This only applies to the first level of elements of alist M ore sophisticated operations on lists must be performed using the S PLUS command language 155 CHAPTER 6 USING THE OBJECT BROWSER Moving and Copying Objects Copying Data from One Database to Another Database Deleting Objects 156 T he data editor for S PLUS data objects is opened by double clicking on the object in the O bject Browser M oreover S PLUS functions can be edited by selecting the Edit menu verb of the function context menu T his activates a script editor containing the function code Finally object names can be changed by selecting the name in either the left or right pane and holding the mouse over the text Asa result an edit control is activated To save the changes press the ENTER key or
54. Formulas Create Formula Click this to open a formula builder dialog used to construct a formula specifying the desired model See chapter 20 Building Formulas Family Select a distribution family for the model Link Select the link function for the model The link function of the response is modeled as the sum of additive terms T he possible link functions depend on the family GENERALIZED ADDITIVE MODELS Save Model Object Variance For the quasi family a variance function can be selected Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed The only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names T he default is last gam T he saved object is of class gam See the on line help for gam obj ect for more information on the saved object Geeemhced Addie Woda E aj Model Optoma Firin Fio Predes Opin cetios Peseta Miaran Henao fat Comenqence Tolerance hon Mmomum Elsckbting Re reticns a Bockhtiag Cormier Tolerance Jiti F Print feretion Trace Options Page Optimization Parameters Maximum Iteration Enter the maximum number of local scoring iterations Convergence Tolerance Enter the converge
55. Gonechon Daia Pisma eects Chan inanon Vensile Ose eae Ae E E Frin fesata Disriioahon veiabka Teas Ohne icstion YeaaHe Thiis Data Frame Select a data frame Classification Variable One Select the factor column that contains the first classification variable T his variable must have exactly two levels Classification Variable Two Select the factor column that contains the second classification variable T his variable must also have exactly two levels Classification Variable Three Select the factor column that contains the third classification variable Apply Continuity Correction Check this if the results to apply a correction for continuity See the on line help for mantel haen test for an algebraic definition of the continuity correction MANTEL HAENSZEL CHI SQUARE TEST Results Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten The saved object will have classht est This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names Print Results Check this to print the results in the designated output window Related S PLus language functions mantelhaen test print htest menuMantel haen
56. Graph To create a graph using the Graph option Using the Menus 1 Choose 2D Plot or 3D Plot from the Graph menu Or if you wish 180 METHODS FOR CREATING A GRAPH to create a multipanel plot of any type choose M ultipanel 2 Choose 2D 3D Matrix Pie or Polar from the Graph Type list Select the desired plot from the Plot Type list 3 In the Graph Sheet field specify the name of the Graph sheet in which you want to insert the graph You can insert the graph on an existing Graph sheet or a new Graph sheet If you specify the name of a Graph sheet that does not exist it will be created automatically Click OK 4 A plot property dialog will appear Specify columns of data to plot on the Data to Plot page Note If you selected data before choosing Insert Graph the graph is generated at this point You can open the plot property dialog by double clicking on the plot 5 Makeany other desired changes to the plot specifications and choose OK 6 If you selected the M ultipane l option a graph property dialog will appear Specify your conditions or other multipanel options here Click OK Creating a Graph See the Programma s Guide for information on creating editable plots using using Scripts or scripts or the Commands window the Commands Window 181 CHAPTER 8 CREATING A GRAPH VIEWING GRAPH SHEETS Zooming a Graph You can zoom your S PLUS Graph sheets to focus on a particular area Sheet
57. Graph sheets using the Insert Picture option You can display an embedded object as an icon in your S PLUS Graph sheet To display an embedded object as an icon 1 Select the embedded object 2 From the Edit menu choose O bject Choose Convert from the submenu 3 In the Object dialog choose the Display as Icon option and choose OK You can specify the formatting and position for the box containing the embedded object You can specify the fill pattern fill color and pattern color in addition to the border line style color and weight To format an embedded object 1 Select the embedded object 163 CHAPTER 7 EXCHANGING OBJECTS WITH OTHER APPLICATIONS Converting to a Different Format Converting to a Picture Importing Graphic Images 164 2 From the Format menu choose Selected O bject A property dialog appears 3 Make any desired formatting changes and choose OK If your Graph sheet contains an embedded object whose source application is not installed on your computer you can convert the object to another format to use with an application you do have installed For example if your Graph sheet contains an embedded table from Word but you do not have Word installed you can convert the embedded table to another word processing format To convert an embedded object to a different format 1 Select the object 2 From the Edit menu choose Object Choose Convert from the submenu 3 Click the C
58. History log if the appropriate History log options were chosen Dialog Rollback lets you scroll through all of these commands as long as they are still in the History log and to view them in the dialog In Dialog Rollback you scroll from the most recent changes back through the older changes S PLUS keeps a continuous record or History log of nearly all operations T heH istory log contains the program commands invoked by each menu and toolbar selection by dialogs and by commands issued in the Commands window The History log can be displayed in a Script window that can be viewed edited and executed You can use the History log to undo erroneous commands by directly editing the command in the script and re executing it H owever this assumes that you have not cleared the H istory log and that you can restore your data to the format it had at the beginning of the script 1 Click the History button on the Standard toolbar or choose History Display from the Window menu The History log is displayed in a script window 2 Edit the History script renoving any erroneous commands for example remove the Remove Column command if you wish to undo acolumn deletion 109 CHAPTER 4 USING DATA WINDOWS 110 3 Click the Run button on the Script toolbar or choose Run from the Script menu The commands in the History log are re executed restoring your D ata windows and Graph sheets to their state prior to the erroneous comman
59. M atrix Plot Axes page 328 Specify the names or column numbers of the variables to use for the scatterplots as the X data You can specify the standard line attributes style color and weight that will be used if you select to draw a smoothed line through the points Click on the D raw Histogram option to have histograms drawn on the diagonal T he standard fill options are available for the histograms fill color pattern pattern color and line style color and weight You can specify all of the options available in 2D line plots for symbol attributes using text as symbols and for varying symbol color sizes and styles according to an additional variable In addition all of the smoothing types available for 2D line plots are available for scatterplot matrices Formatting Matrix Plot Axes To select the axes Click on a horizontal label tick label or tick mark To format the axes 1 Doubleclick on any part of the axes as described above or select PLOT PROPERTIES Display Matrix Graph the axes and choose Selected M atrix Plot Axes from the Format Menu 2 From the M atrix Plot Axes dialog choose the desired tab D isplay Ticks Variable Labels M ake the desired changes and choose OK Axis Display You can specify line style color and line weight for the matrix plot axes Horizontal Vertical Grids You can specify the line style color and line weight for horizontal and or vertical grids T hese will be d
60. MODELS 510 Plot Page Plots Confidence Intervals Curve Options values If Include C ensoring Times is checked the default is the vector of all unique times in the fitted model O therwise the default is the vector of event death times Scaling Factor Enter a number used to rescale the survival time For example if the input data is in days enter 365 25 to rescale to years Model Optom Romme Plot Plots fom Opina F Sure Curve r Leg Soca ery Ceniaences herve Hah Gale kandan E Show Conidae interna z Fack Hte Tick Labuk Cisza Opto fi Latoti f Scat Fehr lv yaaa Tack Labeli Line Tapai fi f Line idhe i ofan La ares preg hanis Y a Label Mark Censonng Timm Geasqng Mark Syrebol ij Saaai PO O O mcs tom LET a Survival Curves Check this to plot survival curves for the current nonparametric survival modal Show Confidence Intervals Check this to plot two sided confidence intervals The confidence interval level can be set on the O ptions page Line Color s Enter a list of integers using either the name of an existing S PLUS vector or a comma delimited list If the number of curves to be plotted is greater than the length of the list the line colors will cycle through the list NONPARAMETRIC SURVIVAL Censoring Marks Axis Options Line Type s Enter alist of integers using either the name of an existing S PLUS vector ora comma delimited list If the number of curves to b
61. Objects with Other Applications Graphics Environment Chapter 8 Creating a Graph Chapter 9 Working with Trellis Graphics Chapter 10 Formatting a Graph Chapter 11 Working with Graph Objects Chapter 12 Editing Plot Properties Chapter 13 Exporting graphs Printing and Sending Mail 15 27 55 85 117 141 159 173 207 227 281 295 337 CONTENTS OVERIVEW Statistical Analysis Chapter 14 Chapter 15 Chapter 16 Chapter 17 Chapter 18 Chapter 19 Chapter 20 Using the Statistics Menus and Dialogs Creating and Manipulating Data Summarizing and Exploring Data Comparing Samples Fitting Statistical Models Using Multivariate Survival and Time Series Models Building Formulas Session Environment Chapter 21 Chapter 22 Chapter 23 Index Working With Script Commands and Report Windows Customizing The User Interface Customizing Your S PLus Session 341 347 361 379 413 487 535 545 567 597 605 CONTENTS Chapter 1 Welcome to S PLus Introduction Installation System Requirements H elp Support and Learning Resources Getting H elp W hat s New in S PLus N ew Features Chapter 2 Working with the Graphical User Interface U ser Interface Using M enus D ialog Boxes and Toolbars U sing the M ouse U sing the K eyboard U sing W indows U sing M ain M enus U sing Shortcut Right Click M enus Specifying O ptions in D ialogs U sing T oolbars and Palettes S PLus Windows O bje
62. Plegeession alle Model Results Frinks Maney Bn VC Shot Guppul F Long si r APIO Taa l Gore tehon bimn of Esimees Printed Results Short Output Check this to display a short summary of the model fit T his includes the model formula the regression coefficients the residual standard error and the degrees of freedom Long Output Check this to display a more detailed summary of the model fit This includes the model formula five number summary of the residuals the coefficients their standard errors t statistics and p values the residual standard error and the degrees of freedom the multiple R Squared value and the F test for the overall model with its degrees of freedom and p value ANOVA Table Check this to display an analysis of variance table T he sums of squares in the table are for the terms added sequentially Type sums of squares Correlation Matrix of Estimates Check this to display the correlation matrix of the regression coefficients This is available only when Long O utput is checked STEPWISE LINEAR REGRESSION Related S PLus language functions Iim step addl dropl menuStep 443 CHAPTER 18 FITTING STATISTICAL MODELS FIXED EFFECTS ANALYSIS OF VARIANCE This dialog performs classical fixed effects analysis of variance and generates plots related to assumptions and results of the procedure The analysis of variance model makes several assumptions about the data for information a
63. SPSS Excel and other file formats DDE OLE client and server graph documents WHAT S NEW IN S PLUS Export graphs as BM P TIFF EPS WMF or other file formats Link data from Excel or other O LE spreadsheets Embed S PLUS graphics in M icrosoft Word and edit in place e Create PowerPoint slides from S PLUS graphics automatically Programming e Validation suite of test cases to confirm accuracy of S PLUS output e OLE Automation client and server Edit and run scripts 25 CHAPTER 1 WELCOME TO SPLUS 26 WORKING WITH THE GRAPHICAL USER INTERFACE Using Menus Dialog Boxes and Toolbars Using the M ouse Using the K eyboard Using Windows Using M ain M enus Using Shortcut Right Click M enus Specifying O ptionsin Dialogs Using Toolbars and Palettes S PLUS Windows O bject Browser D ata W indow Graph Sheet Commands W indow Script Window Report Window S PLUS Menus S PLus Dialogs S PLu s Palettes and Toolbars Standard Toolbar Graph Toolbar D ata Frame Toolbar Script Toolbar O bject Browser Toolbar Report W indow Toolbar Commands W indow Toolbar Plots2D Palette Plots3D Palette Annotation Palette 27 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE User Interface S Plus is a full featured Windows application designed for easy intuitive analysis and visualization of data The Microsoft Office compatible user interface is completely customizable allowing you to tailor the look of S P
64. Scale page has the following options Axis Display Specify the style color and line weight of the lines used to draw the axis Scaling Choose from Linear Log N atural Log In or Probability scaling for the axis Choose Probability to use a scale with data expressed as percentages and a scale range is between 0 001 and 99 999 Data with a true Gaussian distribution sigmoidally shaped curve will plot as a straight line FORMATTING 2D AXES Range Page Hide Choose whether to hide or display the axis Cross Axis Choose whether to have the axis cross another specified axis at the origin Plot s will be drawn in the same location with either option The selected axis will intersect with the axis specified in the Axis to Cross field at the origin if the origin is included within the axes range and all four quadrants will be available for plotting If this option is not chosen the axes will meet in the lower left hand corner regardless of the axes range Axis to Cross Specify the axis number of the x axis or y axis you wish to cross at the origin The axis number appears in the title bar of the property dialog Placement Choose Left Lower to have a y axis snap to the left edge of the plot area An x axis will snap to the bottom edge of the plot area Choose Right U pper to have a y axis snap to the right edge of the plot area An x axis will snap to the upper edge of the plot area Axis O ffset Specify the offset of the axis
65. Specify the tick interval between circular grids and spoke grids Number Size Choose from Numbers Size Auto Column Factor 10 or Factor e Choose Numbers to specify the number of spokes or grids Choose Size to specify the interval size in axes units between spokes or grids Axis Minimum and Maximum Range Specify the minimum and maximum values along the axis Choose Auto to have the minimum and maximum of 259 CHAPTER 10 FORMATTING A GRAPH Labels Page Font Page 260 the radius set equal to the minimum and maximum values in the data plotted The minimum value is ignored and set to 0 if the Data Type specified is Linear Data Type Specify the type of data to be plotted dB decibel Log or Linear In the Polar Axes dialog the Labels page has the following options Label Type See Table 10 1 in the section Formatting 2D Axis Labels page 244 for a complete list of display type options Precision Specify the number of decimal places to display for numeric labels Column Specify the name of a column containing the labels T he first row in the column is used to mark the first major tick mark the second row the second tick mark and so on You can leave any number of rows blank For example if you want labeling to begin at the second tick mark leave the first row in the column blank and enter your labels starting with the second row Data Sheet Specify the name of the Data window containing the label column O
66. T est 406 M antel H aenszel C hi Square T est 408 Pearson s C hi Square T est 410 xi CONTENTS Chapter 18 Fitting Statistical Models 413 Introduction 413 Linear Regression 414 Nonlinear Least Squares Regression 421 Logistic Regression 426 Log linear Regression 427 Robust LTS Regression 428 Local Regression 433 Stepwise Linear Regression 440 Fixed Effects Analysis of Variance 444 Random Effects Analysis of Variance 450 Generalized Linear Models 456 Generalized Additive M odels 465 Tree Models 473 Multiple Comparisons 480 Compare M odes 485 Chapter 19 Using Multivariate Survival and Time Series Models 487 Introduction 487 M ultivariate Analysis of Variance 488 Factor Analysis 493 Principal Components Analysis 500 Nonparametric Survival 506 Cox Proportional H azards 513 Parametric Survival 521 ACF Autocovariance Function 526 ARIMA Modding 528 Chapter 20 Building Formulas 535 Overview 535 Linear Regression 536 Transformation 538 Cox Proportional H azards 541 Chapter 21 Working With Script Commands and Report Windows 545 O verview 545 The Script Window 547 W orking with Scripts 547 xii CONTENTS Using Find and Replace H iding and Unhiding Scripts Context Sensitive H elp Using the Right Button M enu Show Dialog Time Saving Tips for U sing Scripts in S plus H istory Log Dragging Graph O bjects into a Script Window D ragging Function O bjects into a Script Window The Commands Window The Report Windo
67. TIME SERIES MODELS Select a data frame gt Tip Formula Model Save Model Object 514 You can type into the D ata Frame edit box any expression which evaluates to a data frame Weights Enter the column that specifies weights to be applied to all observations used in the regression To weight all rows equally leave this blank Subset Rows with Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the model If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Formula Enter a formula specifying the desired model Create Formula Click this to open a formula builder dialog used to construct a formula specifying the desired model See the chapter
68. a color palette You can use the color palette to edit 16 of your own colors T hese user colors appear in the Color menus for objects in the Graph sheet T he user colors specified when the Graph sheet is created are determined by the Style and Color Scheme definitions in Options To modify user colors for all subsequent Graph sheets you should edit the Color Schemes in O ptions C olor Schemes Number of Colors mage colors are a series of fill colors that can be used for draped surfaces flooded contours and levels plots The specification of image colors consists of up to sixteen core colors and a list defining the number of shades or color gradations between each core color Number of image colors indicates how many core colors are used in the image colors definition Number of Shades A list of numbers separated by commas indicating how many shades should be used between each core color For example if there are three core colors black red and white and number of shades is specified as 5 15 then a total of 23 colors will be used for the image color scheme 5 shades between black and red 15 shades between red and white and the three core colors Image Colors Click on the Image Colors button to access and edit a color palette of the core image colors O nly the number of colors specified by the of Colors prompt will be used in the image color scheme You can save the settings from your Graph sheet so they will be used as defaults w
69. a covariance list as model input instead of a data frame Checking this will enable C ovariance List Covariance List Enter the name of a covariance list to be used as alternative model input T his list must have the form of a list returned by cov wt and cov mve Components must includecenter andcov Acor component will not be used however ann obs Component will be used if present Variables Choose several variables to include in the factor analysis Formula T he Formula edit box is automatically filled using the variables selected from the Variables drop down box T here is no response variable for factor analysis the formula shows the selected variables additively following a tilde The formula field may be edited directly Number of Factors Enter the number of factors to fit T he default is to fit 1 factor Method Select either maximum likelihood mle or principal factor estimation principal T he default is maximum likelihood estimation FACTOR ANALYSIS Rotation Select a rotation to use the default is varimax rotation Save Model Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case se
70. a text box 1 Position the insertion point in the text box If text is highlighted it will be replaced when you begin typing 2 Edit the text Toolbars contain buttons that are shortcuts to menu selections You can use toolbar buttons to perform file operations such as opening a file or saving a file You can also use toolbar buttons to make immediate changes to selected objects such as font or color changes S PLUS displays two toolbars the Standard toolbar and a special toolbar that changes depending on the type of current active window You can turn off the display of Standard and special toolbars but only for the current session When you pause the mouse over a toolbar or palette button helpful Tool Tips small prompt windows will appear You can control whether tool tips are enabled in the G eneral Settings dialog Select O ptions G eneral Settings to see the dialog Statistics Graph Options Window el S osla ey a Restore Data Objects Figure2 4 Using Tool Tips on the Standard toolbar 38 USING MENUS DIALOG BOXES AND TOOLBARS Using Toolbar Buttons Displaying Palette Buttons To select a toolbar button position the mouse pointer over the desired button and click For example you can save your current Graph sheet or script just by clicking on the Save button Or you can open a palette of plot types by choosing one of the 2D or 3D Plots buttons When you position the mouse pointer over a toolbar b
71. amount by which 1932 exceeds 1931 at M orris is similar to the amounts by which 1931 exceeds 1932 at the other sites Either an extraordinary natural event such as disease or a local weather anomaly produced a strange coincidence or the years for M orris were inadvertently reversed M ore Trellis graphics a statistical modeling of the data and some background checks on the experiment led to the conclusion that the data are in error But it was a Trellis graphic such as that created Figure 9 6 that provided the Aha TRELLIS EXAMPLES which led to the conclusion 0 1932 A 1931 Waseca Trebi 7 2 a Wisconsin No 38 No 457 7 0 A Glabron 4 O a Peatland Q 4 Velvet O o KA Manchuria O a 4 0 a Svansota O A Crookston Trebi OA Wisconsin No 38 O Aa No tr 7 oO abron Peatland 8 A Velvet a wed T ae anchuria No 462 0 a Svansota O A Morris Trebi 7 Wisconsin No 38 A 42 No 457 a 0 Glabron 4 40 Peatland Aa 0 Velvet a 0 No 475 a O eed Sal o o J Svansota a O University Farm gt F Trebi O A Wisconsin No 38 No 457 4 Q a Glabron OA Peatland 7 Aa Velvet Lo a meed amp a o 4 Svansota O A Duluth Trebi OA Wisconsin No 38 Oo No 457 O a Glabron OA Peatland lt oa elvet 7 No 475 OA Manchuria A No 462 4 A Svansota 7 a Grand Rapids Trebi a Wisconsin No 38 7 8 A No 457
72. and moved but only within the confines of the main window The restore button replaces the maximize button when the window is maximized The restore button contains two interlocking squares and will return the window to its previous size The menu bar is a list of the available menus Each menu contains a list of commands or actions The scroll bars let you scroll up and down through a window The window border surrounds the entire window You can lengthen or shorten any side of the border by dragging it with the mouse The window corner can be used to drag any two sides of the window The mouse pointer is displayed if you have a mouse installed T he mouse is usually in the form of an arrow an I or a crosshair For more information see the section titled M ous Pointers earlier in this chapter USING MENUS DIALOG BOXES AND TOOLBARS Switching to a Different Window Moving and Sizing Windows At any time you can have many windows open simultaneously in S PLUS The number of windows is limited only by your system s memory resources To switch from one window to another window w Click on any portion of the preferred window that is visible or select the preferred window from the list at the bottom of the Window menu A maximized window cannot be moved or resized A smaller window can be moved or resized within the confines of the application window N ote that not all windows can be resized To move a window or d
73. angle between 0 and 360 degrees and labels are angled counter clockwise accordingly For example if you specify 90 for the angle the tick labels will parallel the vertical axis Minor H eight Ratio Specify the height of the minor ticks as a ratio of the height of the major ticks a value in the range of 0 to 1 FORMATTING AND ROTATING 3D AXES FORMATTING AND ROTATING 3D AXES Formatting 3D Axes To select the axes w Click thex y or z axis to select the axes To format the axes 1 Doubleclick the axes or select the axes and choose Selected 3D Axes from the Format menu 2 From the 3D Axes dialog choose the desired page Display Font Ranges X Text Y Text or Z Text Make the desired changes and choose OK Alternatively you can access pages of the property dialog by selecting the axis right clicking and selecting a page from the shortcut menu Display Font Page n the3D Axes dialog the D isplay Font page has the following options Ranges Page Axis Display Specify the style color and weight of the lines used to draw the axis Tick Position Specify the orientation of the tick marks Specify whether to have ticks drawn from the axis In from the axis O ut to have ticks Cross the axis or not drawn at all O ff Tick Length and Tick Weight Specify the length and weight of the tick marks The length is measured in relative inches cm depending on which you specified Font Options You can specify standard fon
74. are difficult to select in other views such as overlaid graphical elements in a Graph sheet When graph objects are selected in the right pane of the O bject Browser they are also selected in the Graph sheet in which they reside Moreover if a data object is visible in the data editor selecting columns of that object in the right pane of the O bject Browser will select them in the D ata window for that object 157 CHAPTER 6 USING THE OBJECT BROWSER 158 EXCHANGING OBJECTS WITH OTHER APPLICATIONS Overview 159 Embedding O bjects from O ther Applications 161 Creating and Editing Embedded O bjects 162 Importing Graphic Images 164 Linking D ata from O ther Applications 166 Embedding S PLus Graphics Within O ther Applications 168 Updating Embedded Graphs 168 Creating a PowerPoint Presentation 170 Overview S PLUS supports linking and embedding capabilities so you can use data or objects created in other applications in your S PLUS Graph sheets In S PLUS you can link an S PLUS plot to data from another application and have it retain a connection to the source data Plots linked to data are updated automatically when the data changes See the Programmes Guide for more details on DDE and Automation You should link S PLUS plots to data when e the data is likely to change you need the most current version of the data in your S PLUS plot or e the source document is available at all times and updating is necessary You can e
75. builder dialog used to construct a formula specifying the desired model See the chapter Building Formulas for more information T hese fields control the algorithm used by the S PLus function tree to fit the tree models See the on line help fort ree control for details Min No of Obs Before Split Enter the minimum number of observations to include before the first cut on a variable T he default is 5 Min Node Size Enter the minimum node size at which the last split is performed The default of 10 means that growing continues if there are at least 10 observations in anode Min Node Deviance Enter the minimum node deviance before growing stops TREE MODELS Save Model Object Results page Printed Results Saved Results Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed The only non alphanumeric character allowed is the period The default is last tree Tiot Rogimasion Model Arsult Pint Prune Shon Frinks Fna bts Seed Rarua E Bunner Dea open Sve in r Full Tree r ie doeartcedtion Eon P Pearman Fessuels r Detect Ag eich ie Summary Description Check this for a short description of the fitted model Full Tree Check this to print the fitted t
76. by right clicking on the plot and selecting Convert to O bjects from the shortcut menu Page Creation If a command or series of plotting commands are issued additional pages of a Graph sheet can be automatically created If page creation is set to N one the first page will be overwritten when anew plot is created If it is set to Page per Graph a new page will be created for each new plot If it is set to Page Per Command a command consisting of a series of plots will create multiple pages W hen a new high level command is issued from the command line it will overwrite beginning with page 1 If Auto is selected page creation will follow the choice in O ptions Graph This option has no impact on graphs created using the gui Create commands Height Multiplier When using in text codes for super and subscripting each level of super or subscripting is made smaller by this factor numbers ranging from 0 to 4 The default is 0 75 If you enter x 2 in text it would appear as xz N ote that the 2 would be 75 of the size of the x If there was an additional level of superscripting the additional character would be 75 of the size of the 2 Shift Multiplier W hen using in text codes for super and subscripting each level of super or subscripting is raised or lowered by the font height multiplied by the number specified here ranging from 0 to 4 The default is 0 6 The larger the specified shift multiplier number the
77. can create composite graph objects To edit their contents you can convert them into multiple objects 1 Right click on the composite object 2 Select Convert to O bjects from the shortcut menu For a further discussion of the properties of graphic objects such as color and line properties see section Common Plot Properties page 300 M ultiple graphic objects can be edited by 1 Choosing the appropriate formatting options from the toolbars or menus or 2 Selecting the objects and then choosing Selected O bjects from the Format menu To edit multiple objects using the toolbars 1 SHIFT click or CTRL click to select the objects you wish to edit 2 Click on the appropriate toolbar button to change the selected objects to the attribute selected for example the Line Color button To edit multiple objects using the Selected Objects option 1 SHIFT click or CTRL click to select the objects you wish to edit 2 From the Format menu choose Selected O bjects When the dialog appears you can specify values for the properties you wish to change If the selected objects are all of the same type the changes will take effect on all of the objects 3 If you have multiple dissimilar objects selected the dialog will contain properties for only the first object selected H owever modifications made in this dialog will affect the other dissimilar selected objects if they also have the property For example if you have selected se
78. can specify standard symbol attributes style color height line weight and frequency and line attributes style color and weight Curve Fit Type Specify the type of regression curve Linear Exponential Log 10 Log base e or Power The Curve Fit Type and Polynomial Regression O rder can both be specified independently therefore these curve types can be combined For example you can specify a log curve type with a polynomial order of 3 The following table lists the equation that results from the combination of Polynomial with other curve types Table 12 4 Resulting equations from combining polynomials Combining Polynomial with O ther Curve Types Exponential y a exp b x c x 2 Log base 10 y a b log x c log x 2 Log base e y a b ln x c In x 2 Power y a x N b c ln x Omit Constant Choose whether to omit a constant term from the model For example if you choose this option with a linear curve the fitted curve will be forced through the origin PLOT PROPERTIES Results By Confidence Bounds Polynomial Regression Order Specify a positive value here to specify the order of regression of the dependent variable upon the independent variable Any value in the range of 0 to 88 limited by the number of data points is allowed here but values above 9 are likely to result in numerical overflow problems Number of Predicted Values Specify how many predicted values you want S PLUS t
79. can specify the symbol and line color independently in the plot property dialog To change plot properties using the shortcut menu 1 2 Place the mouse cursor on the plot and click the right mouse button The shortcut menu appears to the right or left of the mouse pointe From the shortcut menu choose the desired option to display its dialog 299 CHAPTER 12 EDITING PLOT PROPERTIES Formatting Multiple Plots Common Plot Properties Scaling to Axes 3 Edit the desired properties and choos OK You can format several plots at once using the toolbar buttons To format multiple plots 1 Select the plots you want to change You can do this by SHIFT clicking on each plot or selecting Edit Select All Plots from the menu 2 Choose the desired option from the Graph toolbar For example you can select all your line plots and change them to the same color using the Line Color button on the toolbar S PLUS gives you complete control over all properties of your plots S PLUS uses defaults for all plot attribute options and can automatically alternate or cycle line and symbol styles or fill pattern and color if you have multiple plots within a graph You can reset the cycling defaults at any time by selecting Graph Styles from the O ptions menu Different plot types share some common plot options while other are unique to a specific plot type For example line style color and weight are options for aline
80. can specify the label column for X and multiple Y columns for the Y data specification M ultiple Y columns can be specified in a list for example Y1 Y2 Y3 or in a sequence for example Samplel Sample5 Each column must have the same length See section Preparing D ata for Graphing page 194 for an example of multiple column data Thex and y specifications are reversed for horizontal bar charts Stacked databar charts If your y data are stacked the number of barsin each grouping is equal to the number of rows in column Y divided by the number of rows in column X The first y value will correspond to the first x the second y to the second x and so on repeating throughout the y column See section Preparing D ata for Graphing page 194 for an example of short form stacked data Thex and y specifications are reversed for horizontal bar charts Standard bar chart with error bars For a standard bar chart with error bars specify the names or column numbers of your X and Y data column s You must have more than one Y data column to have vertical error bars computed automatically The mean and standard error are automatically computed and a bar is plotted at the mean If you already have a column containing the size of the error you can specify the column under Z T his column is then used to draw the error bars and the y values are used to determine the height of the standard bars Thex and y data specifications are reversed for hor
81. case sensitive so X and x are different names If there are any columns of classf act or in the data frame then this object is a list with two components each of class t abl e If no columns are of class factor or if factor summaries were not requested then this object is a list with one component If grouping variables are specified the result will be a list of class by with one component for each combination of grouping variables T he structure of each component is determine by whether factors are present as indicated in the previous paragraph Related S Plus language functions by min median mean summary var max print table CONTINGENCY TABLE CONTINGENCY TABLE T his dialog performs a crosstabulation of a collection of factor variables held in columns of a data frame To generate a contingency table Choose Statistics gt Data Summaries gt Crosstabulations from the main menu The dialog shown below appears Contmquncy Table REE Disia Pamo Pemi Subaat Firan wih Eisra i Trest hing wrth hiatal E Fiii Feewelts Cousa Wariabia Zj Forni rs C a a L 1 sslonesabs Data Data Frame Select a data frame Subset Rows with Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped
82. choose one of the Projected 2D options for the Plot Type Choose OK 3 A new Graph sheet and a dialog for the selected plot appear 4 Specify the columns of data to plot on the D ata to Plot page Note If you selected data before choosing Insert Graph the graph is generated at this point 5 Make any desired changes to the plot specifications and chooseO K 6 A 3D graph is drawn with a 2D plot projected onto a default 3D grid plane 7 Double click the grid plane or select the plane and choose Selected Object from the Format menu to change the plane s position or formatting characteristics To project a 2D plot onto an existing 3D graph 1 Click the 2D Plots button to display the plot palette 2 Drag and drop one of the 2D plot buttons onto your 3D graph A 2D plot icon will appear on the graph 3 Doubleclick the plot icon to specify the data columns or drag and drop data columns directly onto the plot icon When the plot is drawn it will be placed on a default 3D plane 202 PROJECTING A 2D PLOT ONTO A 3D PLANE Combining 2D and 3D Plots on One Graph For more information on adding and formatting 3D planes see section Formatting 3D Planes page 251 You can combine 2D and 3D plots on one graph by attaching a 2D plot toa 3D planeon a3D graph For instance you can create a flat map of your 3D surface plot by attaching a 2D contour plot to a3D plane on the same graph To add a projection pla
83. click outside the edit control To cancel any changes press the Esc key Names of objects that are components of lists are not editable Graph objects on the same O bject Browser page are moved or copied by dragging their icons For instance to copy a graph object from one Graph Sheet to another drag the icon and drop it onto the icon for the other Graph Sheet To move the graph object hold down the aLT key while dragging T he same operation can be performed through the objects context menu commands copy and paste or accelerator keys CTRL C and CTRL V Typically a researcher will keep a backup copy of their data in a database that isin a position greater than the working database By doing so the researcher can modify the data in the working database and retain a backup in case any undesirable modifications are made to the working data The databases used by S PLUS are viewable in the Object Browser in a Browser Page that filters on the interface class SearchPath T he single search path object will be found under the SearchPath root node in this Browser Page Expand the contents of the SearchPath root node to view the SearchPath object Select the SearchPath object to view the databases used in S PLUS listed in the right pane The context menu of each D ataBase object has a Paste menu item that will paste objects copied from another database Deleting objects is simple through the O bject Browser this is particularly useful when deleting
84. color and weight and line break options The line style is set to N one for scatter plots 319 CHAPTER 12 EDITING PLOT PROPERTIES Symbol style Vary Symbols 320 Connect Type Choose an option to select a method to connect isolated points Choose Direct to create a standard line plot Choose Isolated Points to allow space between line and symbols Several options allow you to create Step plots Select Vert First to have the first line drawn be vertical Choose H oriz First to have the first line drawn be horizontal Choose H alf Vert First to create a step plot with divided vertical lines Choose H alf H oriz First to create a step plot with divided horizontal lines Choose Vert Only to draw only vertical lines Choose H oriz Only to draw only horizontal lines Choose Draw Horiz Grid Lines to draw horizontal gridlines at major ticks to create a dot plot Other options allow you to create High density plots Choose To X Axis M in to have vertical lines drawn from points to the X axis Choose To Y Axis M in to have horizontal lines drawn from points to the Y axis Choose To X 0 to have vertical lines drawn from points to x 0 Choose To Y 0 to have horizontal lines drawn from points to y 0 For line and scatter plots you can specify standard symbol attributes style color height line weight and frequency Use Text as Symbol C hoose whether to use a user specified string or column of text in any font as plotting symbols
85. data frame specified on the model page will be used to identify the points Plot Partial Residuals Check this to display partial residual plots for all the terms in the moda A partial residual plot is a plot of r Dyxj versus Xj were r is the ordinary residual for the i th observation xj is the i th observation of the k th predictor and b is the regression coefficient estimate for the k th predictor Include Partial Fit Check this to include the partial fit for the term on the plot Include Rugplot Check this to display rugplots on the partial residual plots A rugplot is a sequence of vertical bars along the x axis that mark the observed x values Common Y axis Scale Check this to give all the partial residual plots the same vertical units T his is LINEAR REGRESSION essential for comparing the importance of fitted terms in additive models Model Peni Fia Pred Pirs Chev Ppa n a E Confidence Level i Sere Ir l Fredicime F Dorida lsiera r Sinadand Erin Predict Page New Data Save Enter the name of a data frame to use for computing predictions It must contain the same names as the terms in the right side of the formula for the model If omitted the original data are used for computing predictions Save In Enter the name of an S PLUS data frame in which predictions confidence intervals and standard errors are to be saved If an object with the name you enter does not already exist in
86. data objects for organizing data and computational results T hese are data frames matrices vectors and lists Data Frames The data frame is the most common structure used for data analysis in S PLUS It isatable of datain rows and columns which allows different kinds 87 CHAPTER 4 USING DATA WINDOWS Matrices Vectors Lists 88 of data in the different columns Typically the rows correspond to observations and the columns correspond to variables T he data frame is the preferred structure for storing data The example data frame ky phosis has 81 rows of data on 81 children who have had corrective spinal surgery It has four columns representing the variables Kyphosis Age Number and Start Kyphosis is a two level factor telling whether a postoperative deformity kyphosis is present or absent T he other three variables are numeric vectors Age is the age of the child in months Number is the number of vertebrae involved in the operation Start is the beginning of the range of vertebrae involved in the operation For an extensive discussion of data frames and their uses see the chapter on D ata Frames in the Programmer s Guide M atrices in S PLUS are similar to data frames except that all elements of a matrix must contain data of the same mode Commonly used modes are character numeric complex and logical M atrices can have both row and column names M atrices may be used to store data but the user i
87. dialog boxes that created them except for any spaces See the Programme s G uide for an explanation of the program code To create a dialog in S PLUS follow these steps 1 Identify the S PLUS function which will be called by the dialog T his can be either a built in or a user created function 2 Create the Property objects such as pages group boxes list boxes and check boxes which will populate the dialog 3 Create a Functioninfo object having the same name as the function in sep 1 The Functioninfo object holds the layout information of the dialog associates the values of the Property objects in the dialog with values for the arguments of the S PLUS function and causes the S PLUS function to execute Creating Property objects To create a Property object open the O bject Browser to a page with filtering set to Property Right click on the default object labeled Property in the 577 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE left pane and choose Create Property from the context menu The property dialog shown in figure 22 8 appears Propart Object 1 0 All 23 ype Home Soy Pom Delisi Valea i cpanel F Usa guis Gato g Promet ioue r to Goren Dialig Conii irii T iliy M F ofm Dial Haig String ___j Figure22 8 The property dialog for a Property object To modify the default settings that appear in this property dialog in the future right click on the default obj
88. different formula or use the Create Formula button to bring up a Formula Builder that will build a formula for you See the chapter on the Formula Builders for details M ost dialogs have a Save As field that corresponds to thename of the S PLUS object in which the results of the function call are saved For most dialogs such as those for model fitting the Save As name defaults to last where is related to the type of model In some dialogs the Save As field is blank by default results are not saved M any of the modeling dialogs also have one or more Save In fields The Save In field corresponds to the name of a data frame in which model output is saved Examples include fitted values residuals predictions and standard errors PLOTTING FROM THE STATISTICS DIALOGS Plotting from the Statistics Dialogs Saving Results From an Analysis M ost of the statistics dialogs produce default plots that are appropriate for the analysis M any have several plot options usually on a separate Plot tab By default these plots are created using traditional S PLUS graphics and as such they are not editable To create editable plots right click on the data part of the graph and choose Convert to O bjects from the context menu A global option to create editable graphics by default can be set via the O ptions menu Select Options Graph Options and check the Create Editable O bjects box Editable graphics however generally take longer to
89. drawn The plane is automatically positioned at the minimum or maximum position on the plane or 1 from the Insert menu choose 3D Planes 2 From the3D Planes submenu choose the desired plane 3 Make any desired formatting specifications in the 3D Planes dialog and choose OK 251 CHAPTER 10 FORMATTING A GRAPH Changing the Plane Position Page Grids Page 252 In S PLUS you can switch between different 3D planes using the plot palettes or the 3D Plane dialog If any 2D plots are projected onto the plane they will move with the plane To change the plane using the palette 1 Select the 3D plane you wish to change 2 Click the 3D Plots button on the standard toolbar A palette of 3D plot types and 3D planes appears 3 Click the desired 3D plane The selected plane is redrawn using the new plane To format a 3D plane w Doubleclick the 3D plane or w Select the plane and choose Selected 3D Plane from the Format menu In the 3D Plane dialog the Position page has the following options Projection Plane Specify whether to draw the grid on the XY XZ or YZ plane Position Specify the position of the projection plane within the 3D workbox in axes units For example if you specify 1 for the position of the XY plane the XY plane will be placed at the position where z 1 Enter M in to have the plane placed at the axis minimum or M ax to place it at the axis maximum Draw in Panel s For a multipanel graph
90. easily select data functions and objects to simplify the preparation of your analysis Figure2 5 An Object Browser window several can be open Smultaneougy Data Window 40 The Data window displays data sets in an editable spreadsheet format It handles data in a column oriented manner Data can be edited within the D ata window Columns can also be copied or moved from one D ata window to another T his allows you to easily manipulate your data for a wide variety of operations and analyses In look and feel the S PLUS Data window is very similar to standard spreadsheets H owever the D ata window has additional features which make it easy for you to create a graph explore your data and perform advanced S PLUS WINDOWS i kyphosis Pos GP fe a a l statistical analysis and modeling S PLUS imports and exports data in all popular formats including SAS SPSS Excel and ASCII as well as from any ODBC compliant application O nce selected in the D ata window columns of data may be plotted simply by clicking one of the many 2D and 3D plot palette buttons F100 3 00 5 00 i 156 00 3 00 14 00 126 00 4 00 5 00 5 00 1 00 4 00 15 00 2 00 16 00 2 00 17 00 T m omm am omm Lad Figure2 6 TheData window with one column variable highlighted Graph Sheet Every plot of data in S PLUS is displayed in a fully customizable Graph sheet Each Graph sheet can contain one or more graphs and you can work wit
91. errors 551 find and replace 552 hiding 555 hiding and unhiding 554 printing 550 running 549 saving 549 stopping 551 undoing edits 552 unhiding 555 using history log 558 selecting columns 90 selecting rows 91 selective undos using dialog rollback 108 setting defaults saving defaults See setup exe 16 shading range 302 short form stacked data 198 Show Dialog option 556 smart cursor 598 Smith charts 326 smooth curves 372 smoothing 371 data 321 smoothing parameter 378 Snap to Grid 601 S news mailing list 20 sorting 106 special colors 302 Spin Window 206 spline smoother 377 S Plus objects 145 S Press newsletter 21 SQL queries 133 stacked 3D contour 330 stacking pie chart labels 326 standard bar charts data specifications 304 standard deviation 363 standard errors 420 start angle pie charts 325 INDEX start column when importing 121 start row when importing 122 Statistics menu 341 StatLib 20 stepwise linear regression 440 sum of all numeric values 363 summaries of the categorical variables 363 summary statistics 362 supersmoother 372 surface plots 332 color range 334 fill color 333 fill surface 333 fill type 333 gridded matrix data 197 increment multiple 334 long form data 199 number of colors 333 plot options 332 special colors 334 stacked data 197 survival curves 518 symbol color 301 formatting 320 frequency 302 height 302 style 301 using text 320 weight 302 symbol style 301
92. existing data frame T he other is to get densty probability or quantile using a sequence of numbers 355 CHAPTER 15 CREATING AND MANIPULATING DATA Data Probability or Quantile Choose Use D ata Frame when the input is from a data frame C hoose Create Sequence to create a sequence of numbers with equal spaces Data Frame Select a data frame from the drop down box Its columns will be displayed in the Variable drop down box Variable Select the column of the data frame for which density cumulative probability or quantile are calculated Starting Value Enter the starting value of the sequence Ending Value Enter the ending value of the sequence Number of values Enter the number of points in the sequence Density Check here to calculate the densities corresponding to the values of the column chosen in Variable or the points of the sequence Cumulative Probability Check here to calculate cumulative probabilities corresponding to the values of the column chosen in Variable or the points in the sequence Quantile Check here to calculate the quantiles corresponding to the values of the column chosen in Variable or the points in the sequence gt Tip When the data contain values of a random variable check D ensty and or Cumulative Probability When the data are probability values check Q uantile to compute quantiles that are random variates When Quantileis checked both D ensty and Cumulative P
93. frame Weights Enter the column that specifies weights to be applied to all observations used in the linear regression To weigh all rows equally leave this blank Subset Rows with Enter an S PLUS expression which identifies the rows to use in the analysis ROBUST LTS REGRESSION Formula Save Model Object To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the model If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Formula Enter a formula specifying the desired model The formula specifies which regression model is to be fit In its simplest form a formula consists of the response variable a tilde and a list of predictor variables separated by s An intercept is automatically included by default For example Fuel Weight Di
94. higher and lower the super and subscripts Font 1 2 3 4 Specify the four default fonts to be used with in text codes in title legend comment and date stamp text In text codes allow you to change the font color and super and subscripting within a line of text Auto Redraw Plots C hoos this option to have plots redrawn automatically whenever changes are made If this option is not chosen plots will only redraw when you choose Redraw Plots Now from the View menu This option is useful when you want to make multiple changes without waiting for plots to redraw after each change T he Auto Redraw Plots option can also be toggled on and off from the View menu FORMATTING A GRAPH SHEET Colors Page Saving Graph Sheet Defaults Background C olor Specify the background color for your Graph sheet Print Background Specify whether you want the background color to print when you print your Graph sheet You may want to turn this option on when creating slides or using a color printer Edit Colors Click the Edit Colors button to access the Color Palette You can use the Color Palette to define 16 of your own colors Custom colors will then appear in the Color menus for the current Graph sheet You can save these custom colors to be used for all subsequent Graph sheets by right clicking in the Graph sheet and choosing Save Graph Sheet Properties as D efault from the shortcut menu User Colors Click the User Colors button to access
95. holding down the mouse button on menu options and when moving the mouse over toolbar buttons the status bar at the bottom of the window describes the function of the menu option or button Clicking Pointing at an object and quickly pressing and releasing the left mouse button Some tasks in S PLUS require a double click quickly pressing and releasing the left mouse button twice Right Clicking Pointing at a selected object and quickly pressing and releasing the right mouse button This brings up a shortcut menu for the Selected object Dragging Pointing at the object then holding down the left mouse button while moving the mouse Releasing the left mouse button drops the object in the new location T he mouse pointer changes shape to indicate what action is taking place T he following table shows the different mouse pointer shapes and the significance of each 29 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE Table 2 1 Different shapes of the mous pointe Pointer M ouse Action S ke Pel i El e Le Gel H l 2 Using the Keyboard 30 Selection mouse pointer Text indicator slanted pointer indicates italic text Crosshair for precise positioning of graph objects The symbol to the lower right shows the object to be drawn a filled circle for example Select a D ata window cell or drag a block of cells Column width indicator appears when mouse is positioned between two columns
96. if the X and Y columns contain character data S PLUS assumes that the grid values in the X and Y direction are a series of integers If character data are specified they will be used to label the X and Y tick marks If you specify the number of data grids the number of X grids times the number of Y grids must be equal to the number of rowsin theZ column For example if your Z column contains 875 values it might represent a grid with 35 X values the number of X grids and 25 Y values the number of Y grids Short form stacked data TheX and Y data can be specified in a short or long form In the short form each X and Y value is listed only once Both columns must have values that arein ascending order T he Z values do not correspond to the X and Y values in the same row S PLUS assumes that the X values vary faster in determining the X Y coordinates that correspond to each Z value For example Table 8 4 represents a list containing vectors X Y and Z Table 8 4 Sample tacked data short form X Y Z 1 2 10 0 60 2 2 20 0 67 3 2 30 0 67 4 2 0 69 5 2 0 71 0 59 0 62 0 65 0 66 0 67 PREPARING DATA FOR GRAPHING Table 8 4 Sample tacked data short form Long form stacked data In the long form the X Y and Z columns are of equal length and can be stored in a data frame T hey contain every combination of the X and Y values and their corresponding Z value The X values are an ascending sequence and the s
97. interactions type gt fit int lt Im ozone wind temperature radiation 2 environmental 2 Fora brief summary of the fit gt fit int 3 For a detailed summary gt summary fit int 4 For an F test comparing this model to the model we fit through the dialogs gt anova last Im fit int One of the major benefits of having a model object is that we may then obtain quantities such as residuals fitted values and predicted values at will We can use standard S language functions to produce a plot of the density of the residuals as one way to assess the normality of the residuals W hen we are ready to edit the plot we will transform it into an editable object 1 Extract the residuals gt resid int lt resid fit int 2 Compute and plot the density estimate gt plot density resid int type line Wewill edit our density plot 1 Right click on the density line Choose Convert to O bjects from the menu 2 Select the X axis title with a single click Then click it again to activate the in place editor Replace the label with Residuals Click outside of the in place editor when you are finished 3 Click on the plotted density Use the Line Color button on the toolbar to change the line color to M agenta IN DEPTH TOUR Creating In S PLUS you can automatically create a PowerPoint 7 0 presentation from PowerPoint your graphs We will use the graphs we have created during this demonst
98. labeled with the car description 60 QUICK TOUR 3 Close the Annotations palette Annotation Figure 3 7 TheLabe Point button in the Annotations toolbar Editing the Graph Now imagine that you wanted to include your graph in a report or presentation You might want to modify its attributes further In S PLUS the graphics are object oriented which means you have complete control over every detail You can easily modify graph objects using shortcut menus dialogs or the toolbar 1 Shift click on each axis title on the graph to select them Change the font size to 20 using the graph toolbar Click on any data point to select the plots A single green square at the bottom center will appear Change the plot color to Red using the drop down Color menu by clicking on the Line Color button on the graph toolbar see figure 3 8 Figure 3 8 TheLineColor button on the Graph toolbar changes text aswell asline colors Creating a 3D Graph S PLUS offers a variety of 3D plot types for powerful data visualization First let s select some 3D data Before we begin close all of your Graph sheets and D ata windows and open the O bject Browser 1 Use the Find button on the Object Browser toolbar to find the galaxy data frame Go to the O bject Browser and click once on the galaxy data frame in the left pane T he columns of this data frame will appear in the right pane Select the data columns east w
99. menu By either method the O bject Browser property dialog is displayed with the Browser tab active An example is seen in Figure 6 3 A description of the properties of the Browser tab of the O bject Browser dialog are listed below Name Enter a name for the O bject Browser window T his name will appear at the top of the window as seen in Figure 6 3 Description Enter a convenient description Bitmap Tab Bar Check this to display icons on the page tabs Leave this box unchecked to label the tabs with text File T his field displays the external file if any to which the browser has been saved CUSTOMIZNG THE OBJECT BROWSER Ubyt Brewer Brewin Fight Pare des k esm E Bitty Tati b pe pe s Pfs BE Figure6 3 TheBrowse tab of the O bject Browse property dialog Customizing the Right Pane Display The right Pane tab of the O bject Browser dialog is shown in figure 6 4 The properties it contains are described below View Select the radio button giving the desired view of objects in the right pane Page Size Enter the number of items which can be accessed by scrolling in a right pane page Some databases contain more than 1000 objects Restricting the number of objects displayed in the right pane list improves display performance When a large number of items must be viewed in the right pane they are organized on different pages within the pane Use the paging buttons Ato navigate betw
100. multiple objects of the same type since like objects are grouped together in the hierarchy For instance when deleting all the arrows on a plot it is much easier to select them in the O bject Browser than directly on the graph Hold down the cTRL key to select multiple discontinuous objects and use SHIFT click to select all objects in a block Press D elete to remove the selected EDITING OBJECTS AND DATA MANIPULATION objects Note The O bject Browser can delete objects only from D atabase 1 T his prevents you from accidentally deleting system functions and data sets Modifying Object Properties Selecting Objects O bjects can be modified in the browser by accessing their properties dialog either by double clicking on an object to open its property dialog or right clicking on an object and selecting Properties from the shortcut menu Changes made through the O bject Browser are reflected immediately in the object Different behavior can be observed when double clicking on objects of different classes For example the D ata window for S PLUS objects such as data frames matrices or vector objects is initiated by double clicking on the object in the O bject Browser M oreover Class nfo objects define the double click action for objects of specific classes Typically the double click action defined in a ClassInfo object executes an S PLUS function The O bject Browser can be used to select objects that
101. new object or can be initialized from an existing file The embedded object becomes part of the S PLUS Graph sheet increasing the file size of the entire Graph sheet Some applications support in place activation meaning that when you double click on the embedded object to edit it the S PLUS Graph sheet remains visible around the object When you embed an object from an application that supports in place activation a hatched border surrounds the embedded object and the application name in the title bar changes from S PLUS to the name of the object s source application In addition all the menus except File and Window change to use the source application s menus and the toolbars change to use the source application s toolbars To embed a new object 1 With a Graph sheet in focus choose O bject from the Insert menu 2 To create and embed a new object choose the Create N ew button and select the type of object Choose OK N ow you can create and activate the new object 3 When you are finished editing the object click outside the object to deactivate it T he embedded object is displayed in your Graph sheet To edit the embedded object in place 1 In your S PLUS Graph sheet double click the embedded object or Select the object choose O bject from the Edit menu then choose Edit 2 The embedded object remains in the S PLUS Graph sheet but the menus and toolbars change to those of the source application Edit the object u
102. of the line For error bars the selection knobs appear at the ends of each bar To select a single object w Click on the object with the mouse pointer Green selection knobs appear on the outline around the object If they do not appear the object is not selected Try to select it again by clicking on the Selection arrow tool on the Annotation palette placing the mouse pointer over the object and clicking again To select multiple objects 1 Click on the object 2 Hold down the SHIFT key or CTRL key and click on another object 3 Repeat step 2 until all the desired objects have been selected or 1 Position the mouse pointer in one corner of the area surrounding the objects you wish to select 2 Click and drag until the selection box encompasses all the objects you want to select O bjects must be entirely within the selection box SELECTING OBJECTS in order to be selected Note If you press CTRL while dragging the selection box will select everything it touches as well as objects inside the box W hen the mouse button is released all the objects within the selection box have green selection knobs displayed on their outline To cancel the selection just click the mouse pointer in a blank area of the Graph sheet Selecting multiple objects is convenient when you wish to move or edit a number of objects To select all objects All drawing objects can be selected in two ways 1 From the Edit menu
103. on Building Formulas for more information Type of Censoring Select the type of censoring right left or counting Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of COX PROPORTIONAL HAZARDS Options Page Optimization Parameters alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to aname that starts with last For example last coxph is the most recent Cox modal fit Coe Pra netinnal Heres 2s p B Moda Ophora Freaks Practica Burvivel Cotes Opteron Pesman Modal Opin Comeangance Tolerance ction d to Tiers anne osm P Robani variana Erima inis Paramon waagt Maimun hernion iD F Alice bolnem Convergence Tolerance Enter a number specifying the convergence tolerance Iteration will continue until the relative change in the log likelihood is less than this number Initial Parameter Values Enter a vector of initial values If this is left blank zero will be used for each variable Maximum Iteration Ente a number specifying the maximum number of iterations If convergence has not been reached after this number of iterations the procedure will
104. on a Graph sheet Using Cut O bjects can also be moved by cutting or copying them and pasting them ina Copy and new position When you cut or copy an object it is placed on the Clipboard and can be placed in another Graph sheet or another application Paste Options To move or copy objects using the menus 1 Select the object or objects you want to move or copy If more than one object is selected the objects will keep the same relative positions when placed in their new location 288 MOVING AND COPYING OBJECTS 2 Choose either the Cut or Copy button from the standard toolbar or from the Edit menu choose the Cut or Copy command 3 Position the mouse pointer in the target Graph sheet or another application for example Word 4 Choose the Paste button from the standard toolbar or from the Edit menu choose Paste or click on the right mouse button and choose Paste from the shortcut menu Sizing Objects D rawn objects can be sized by dragging the green selection knobs that appear on the object when it is selected To resize objects 1 Select the object or objects you wish to size 2 Position the mouse pointer over a knob 3 Click and drag the handle to draw the object to the size you wish A dashed outline traces the new dimensions of the object To size an object horizontally or vertically drag the green selection knobs on the sides of the object To size an object in both dimensions at once drag the gr
105. on the dialog fields follow ODBC Data Source Once you have installed the ODBC Administrator and drivers that connect to the external database you may need to configure one or more ODBC D ata Sources if one has not already been set up for the data source you wish to connect to A Data Source consists of the data you wish to access the application that has the data and the computer and network connections necessary to reach the data Configuring this can be done either using the ODBC applet from the control panel or within S PLUS by selecting the button labeled in the ODBC Import and ODBC Export dialogs If this applet does not exist or selecting the button fails you need to refer to the section Installing ODBC page 130 for details on installing the ODBC Administrator ODBC If you are running ODBC Administrator 3 0 you can then select a tab that corresponds with the type of DSN Data Source Name you wish to create User DSN System DSN or File DSN The type of DSN controls access to the D ata Source you are creating Specific descriptions of the DSN types can be found on each page of the applet After selecting the proper tab select the Add button then select the driver that will export the data from the database to ODBC and select the Finish button If the list of drivers is enpty or does not contain a driver for your database or application you need to install the database or its O DBC driver Contact
106. one is selected the fill type specifications for each contour will be ignored Fill Type You can choose a solid color for each contour image colors to use the color scheme specified on the colors page of the Graph sheet dialog fill the contour plot with a series of shades determined by a range of colors or specify new colors for the filled contours using the standard color dialog See the information on fills in the section Common Plot Properties page 300 for details on fill and color attributes Specify the border and fill attributes for each contour You can make color or style changes to all contours at once or specify properties for individual contours You can choose from the following types of curve fittings 1 Linear 313 CHAPTER 12 EDITING PLOT PROPERTIES Data to Plot Line Symbol Curve Fitting 314 Polynomial Log Base 10 Log Basee Ge E Exponential 6 Power S PLUS performs a standard linear regression displaying a regression line with a scatter plot of the associated data points Regression lines are generated using an ordinary least squares O LS analysis to calculate y values for given values of x For curve fits all statistics are adjusted appropriately For curve fitting plots specify the names or column numbers of your X column and Y column s The X column should contain the dependent variable The Y column should contain the independent variable regressor For curve fitting plots you
107. options below Successes Variable Select the variable that contains the observed counts of successes This column must contain only nonnegative integers T his could also be a vector or an S PLUS expression that evaluates to a vector of integers Trials Variable Select the variable that contains the corresponding number of trials This column must contain only positive integers T his could also be a vector or an S PLUS expression that evaluates to a vector of integers Hyp p Variable Select the variable that contains the hypothesized values for p the probabilities of success T his is not required and it will default to 0 5 for one group If more than one group theN ull hypothesis is that the true probability of success p isthe same for all groups Alternative Hypoth 402 PROPORTIONS TEST Results Select the alternative hypothesis two sided trueparameter is not equal to p default e less true parameter is smaller than p greater true parameter is greater than p Confidence Level Enter the confidence level desired for the returned confidence interval T he default of 0 95 yields a 95 confidence interval Apply Yates Continuity Correction Check this to correct for continuity See the on line help for prop test for an algebraic definition of the continuity correction Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents wil
108. or by the number of X and Y values If the X and Y data specifications are left blank and the data are in one long column the number of X and Y grids corresponding to the Z data must be specified here Output Grids Specify the desired number of X and Y grids to be displayed on the surface If the number of output grids exceeds the number of data grids a 3D spline will be used X and Y Minimum and Maximum Specify the range of X and Y for which you want the surface drawn When Auto is specified the minimum and maximum values in the source X and Y columns are used Z Minimum and Maximum Specify the range of Z for which you want the surface drawn W hen Auto is specified the minimum and maximum values in the source Z column are used Specify a smaller range of Z to draw a subsample of the data Grid D ata Specify No if your data columns are already gridded Specify Yes if your data columns are irregular If you choose Auto S PLUS will determine if the data are gridded or irregular If the data are irregular S PLUS will grid the data automatically Algorithm Bivariate uses the interp function to compute a bivariate interpolation for irregular grids T he triangulation scheme used by interp works well if x and y have similar scales but will appear stretched if they have very different scales The spreads of x and y must be within four orders of magnitude of each other fori nt er p to work It uses a cell search algorithm internal to the gra
109. or issue the corresponding command depending upon the type of dialog O ne use of Dialog Rollback is to restore an object to a previous state This is different from undo in that rollback can be applied selectively to individual objects Typing and The following tasks can be performed in dialog boxes using the special keys Editing in Dialog listed below Boxes Figure2 3 Shortcut keys when using dialog boxes Action Special Keys M ove to the next option in the dialog Tab M ove to the previous option in the dialog Shift T ab M ove between pages in a multi page dialog CTRL Hab M ove to a specific option and select it ALT tunderlined letter in the option name Press again to move to additional options with the same underlined letter D isplay a drop down list ALT DOWN direction key Select an item from a list UP or DOWN direction keys to move ALT DOWN direction key to close the list Close alist without selecting any items ALT DOWN direction key Many dialogs contain text edit boxes Text boxes allow you to type in information such as a file name or a graph title 37 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE Using Toolbars and Palettes Tool Tips To replace text in a dialog 1 Select the existing text with the mouse or press aLT underlined letter in the option name 2 Type the new text Any highlighted text is immediately overwritten when you begin typing the new text To edit text in
110. original data will be overwritten Thus it is usually a good idea to specify a target data column so you always retain a copy of your original data N ote also that if you sort in place and you sort only some of the columns in a data set these columns may be mismatched with respect to the unsorted columns If more than one column is being sorted the sorted columns are placed starting at the specified target column D ata entered after sorting must be entered in the sorted order Powerful sorting capabilities are also available using S PLUS s functions from within the Commands window 107 CHAPTER 4 USNG DATA WINDOWS UNDOING ACTIONS To Undo Actions You can undo most changes to objects For data objects you can usually and Changes undo your last change The same is true for text changes in the Script window For most other objects multiple undo s are possible Figure 4 10 The Undo Undo Lis Redo Redo Lig and Restore D ata Object buttons on the Standard toolbar To undo the most recent action w Click the Undo button on the Standard toolbar or from the Edit menu choose Undo or press the hot key cTRL4z To undo multiple actions w Click on the Undo List button on the Standard toolbar and select the action from which you want to start undoing There is no multiple undo available for data objects H owever the Restore D ata O bject option on the Standard toolbar allows you to restore the object to its initial st
111. partial autocorrelation to estimate the partial autocorrelation Change Maximum Lag Default Check this to give a value for the maximum number of lags at which estimates will be calculated Maximum Lag Enter the desired maximum number of lags at which to estimate the autocovariance or autocorrelation function Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names T he default is last acf This will be saved as a list object See the on line help for acf for more details on the returned object Plot ACF Results Check this to display a plot of the estimates of covariance or correlation against their corresponding lags A 95 confidence interval around the zero line will be included Related S PLus language functions acf lag acf plot ar menuAcf 527 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS ARIMA MODELING This dialog fits univariate Autoregressive Integrated Moving Average ARIM A models estimated by gaussian maximum likelihood See the on line help for ari ma ml e for details on the algorithm and returned values To fit an ARI
112. plot while bar type base and grouping are options for a bar chart The following topics are plot attributes common to many plot types in S PLUS which appear on various pages of the plot property dialogs Variations from the options shown below will be discussed in the specific sections for each plot type X Axis Y Axis f you have multiple axes pairs on your graph use the Scale to options to choose which x axis and y axis to be use for scaling the plot Specify the number of the desired x axis and y axis By default S PLUS scales the plot to the first x axis and y axis on the graph Note it automatically If you specify an axis number that does not exist it will be added automatically For example if you specify 2 for the Y axis and you currently have only one y axis a second y axis is added and the plot is scaled to 300 T his option is available only for 2D plots all 3D plots are scaled to only one set of available axes X Y and Z PLOT PROPERTIES Cropping Hiding For information on adding axes to a graph see the chapter Formatting a Graph Plane M any of the 2D plot types can be projected onto a 3D plane To project a2D plot put the 2D plot on a3D graph and specify the projection plane next to Plane For information on projecting 2D plots see section Projecting a 2D Plot onto a 3D Plane page 201 Check the Crop option to have your 2D plots cropped at the edges of the plot area Leave t
113. represents the variable whose label appears below it The non highlighted points make up the BRUSH AND SPIN Searching for Cases Brush Symbols and Size Control More Brushing Options positive mass in the histogram The probability mass representing highlighted points appears as negative mass that is points downward on the histogram plot 4 You can brush all the points representing one bar in any histogram plot M ove the mouse cursor over a bar in a histogram N ote that only the mouse cursor not the brush appears in the histogram plots 5 Click the left mouse button to highlight all the data points represented in that histogram 6 Click on the right mouse button on the bar to downlight this point You can access the data by its row name or number To the right of the scatter plots thereis alist box containing the row names or numbers 1 Left click on a name in the list box to highlight the points representing that row 2 Click again to downlight Above the rectangle containing the case names is a row of four highlighting symbols A small square is around your current highlighting symbol 1 Click on another symbol to change the highlighting symbol 2 To adjust the size and shape of the brush move the mouse cursor into the background area of the Brush window then press down on the left mouse button The brush appears with the mouse cursor pointing at the lower right corner 3 While holding
114. sheet Exporting to Two options for exporting graphs to EPS files Encapsulated PostScript Files e The first EPS option in the export list uses a traditional export filter and includes a TIFF representation of the graph It is listed as EPS w TIF Header EPS The TIFF representation allows you to see the graph when you place it in other Windows applications but will increase the file size substantially The TIFF representation is used only for screen viewing the graph will be printed using Encapsulated PostScript resolution e The second EPS export option in the export list uses a traditional export filter and is listed as EPS EPS It does not include a TIFF representation of the graph so it produces smaller EPS files Restrictions You can print only one page of a multi page document at atime 340 USING THE STATISTICS MENUS AND DIALOGS Introduction to Statistics Menus and Dialogs Introduction to Statistics M enus and Dialogs 341 Dialog Fields 342 Plotting from the Statistics Dialogs 343 Saving Results From an Analysis 343 S PLU S Functions Called by Statistics D ialogs 344 M odifying the Statistics D ialogs 345 M uch of the statistical functionality of S PLUS can be accessed through the Data and Statistics options on the main menu The Data menu includes dialogs for tabulating data calculating distribution functions and generating random samples and random numbers T he Statistics menu includes dialogs fo
115. stop Allow Collinearity Check this to allow for collinearity in the model matrix columns that are linear combinations of earlier columns will be skipped C oefficients for such columns will be missing NA and the variance matrix will contain zeroes 515 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS 516 Model Options Results Page M issing coefficients are treated as zeros for ancillary calculations Method for Ties Select a method for handling ties efron breslow and exact The exact method computes the exact partial likelihood which is equivalent to a conditional logistic model Robust Variance Estimate Check this to calculate a robust variance estimate T his is the default if the model as defined in the Formula field contains a cluster operative Coe Pra netinnal Hiweneds E lt mE F Moda Options Fear ute Predici Eurial Curd Fririari Diam ie l hotii F Long Quipat De a ea p me Printed Results Short Output Check this to print a short summary of the fit Long Output Check this to print a long summary of the fit T his includes the call model coefficients standards errors of the estimates a confidence interval for the relative risk of each coefficient a likelihood ratio test Wald s test and the efficient score test COX PROPORTIONAL HAZARDS ne Pra ortinnal Hazards p E mE Moda Opiors Fearauts Pretec Survival Caras hike Diin e Type in
116. summary Im Other related S PLus language functions aov gam glm loess nls NONLINEAR LEAST SQUARES REGRESSION NONLINEAR LEAST SQUARES REGRESSION T his dialog fits nonlinear regression models via least squares To perform nonlinear regression Choose Statistics gt Regression gt Non Linear from the main menu T he dialog shown below appears Hoalinoer Lees Squats Aegressice Bl Mndei Cpiow Penta Predis e Dem Sawa Moke Obid Daia Frame Ea Am asish Modal Front Pesmeen nee ue Model page Data Data Frame Select a data frame containing the data for the nonlinear regression gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame Model Formula Enter an expression in the S PLUS language specifying the nonlinear regression model The variables used in the formula are the columns of the data frame the parameters to be estimated and S PLUS functions The response variable appears first followed by a tilde and then the function to be minimized For example y bO exp bl x 421 CHAPTER 18 FITTING STATISTICAL MODELS Save Model Object 422 Options page Optimization Parameters For details on specifying formulas see the chapter on Nonlinear M odes in the Guide to Statistics Parameters name value Enter a comma separated list of the parameters in the formula that are to be estimated along with their initi
117. surface plots Table 8 3 A data frameisan example multiple column data set Levels X Samplel Sample2 Sample3 Sample4 Sampled High 1 0 45 0 69 0 66 0 66 0 19 M edium 2 0 89 0 12 0 41 0 89 0 78 Low 3 0 42 0 61 0 37 0 29 0 44 Specifying Matrix In a gridded surface or contour plot the Z values represent the height of each Form Data intersection in a grid If the Z data are specified in a series of columns in a data frame or in a matrix the dimensions of the grid are defined by the number of rows X grids and columns Y grids of the Z data For example if your Z data consists of 4 columns each containing 5 Z values the number of X grids is 5 and the number of Y gridsis 4 Specifying In a gridded surface or contour plot the Z values represent the height of each Stacked Form intersection in a grid If the Z data are stacked in one long column S PLUS Data will need more information to determine the dimensions of the grid You can do this by specifying either the X and Y data columns if the number of X and Y D ata Grids are set to Auto or the number of data grids 197 CHAPTER 8 CREATING A GRAPH 198 If you specify the X and Y data columns S PLUS uses the minimum and maximum data values to determine the position of the contour along the x and y axis For example if the X column has a minimum value of 2 and a maximum value of 9 the plot will be drawn between 2 and 9 on the x axis If X and Y are not specified or
118. the Edit menu CTRL V The Script Window has its own Undo capability which is separate from the Undo used while you work with Graph sheets and D ata windows W hile you edit a script you cannot undo or redo any actions for Graph sheets and data objects W hile editing scripts you can undo your typing changes by choosing Edit Undo from the menus As soon as you leave the Script window your Undo queue for Graph sheets or D ata windows is restored To undo the last change made in a Script window w Click the Undo button on the standard toolbar or choose Undo from the Edit menu or typecTRL Z T he last change you made will be undone If you need to restore the Script window to its previous state before your last undo you can Undo again and the change you just undid will be restored To review or change text in a Script window use the Find or Replace options You can use Find to locate specific occurrences of text in your script You can use Replace to locate the text and replace it throughout your script Find and Replace can be used for certain words phrases or sequences of characters such as whole commands S PLUS will replace specified text throughout a script unless you select a part of the script It is a good idea to save your script before you use Replace so that if you do not like the results you can close the Script window without saving the changes You can also use U ndo to undo the last replacement made to the script
119. the file type In addition because FASCII files are assumed to be non delimited e g there are no commas or spaces separating fields you also need to specify each column s field width and data type in the Format String This tals S PLUS where to separate the columns Each column must be listed along with its data type character or numeric and its field width If you want to name the columns specify a list of names in the Column N ames field Column names cannot be read from the FASCII data file When importing FASCII files you need to specify the following options in the O ptions page Column Names Enter a list of column names for the imported data columns separated by spaces or commas Specify one column name for each imported column e g Apple O ranges Pears You can use an asterisk to denote a missing name e g Apples Pears Format String Specify the data types and field widths of the imported columns For each column you need to specify a sign then the field width and then the data type Commas or spaces must separate each specification in the string T he format string is necessary because formatted ASCII files do not have NOTES ON IMPORTING FILES delimiters such as commas or spaces separating each column of data H ere isan example format string 10S 12f 5 10F The numbers denote the column widths s denotes a string data type f denotes a float data type and the asterisk denotes
120. the plot and choose data to plot from the shortcut menu 2 On the Data to Plot page choose the H ide option and choose OK The plot appears in its iconized form in the upper left hand corner of the graph T he iconized plot will not appear in the printed output To unhide a plot 1 Doubleclick the iconized plot or select the iconized plot and choose Selected Plot from the Format menu or right click on the plot and choose D ata to Plot from the shortcut menu 2 On the Data to Plot page choose the H ide option and choose OK To delete a plot 1 Select the plot to be deleted If the plot is difficult to select you may want to use the O bject Browser to select it 2 Pressthe DELETE key or from the Edit menu select C lear EDITING DATA SPECIFICATIONS EDITING DATA SPECIFICATIONS Adding or Replacing Data in the Plot You can drop data onto a plot icon or directly on a plot When you drag the mouse directly over a plot icon or a plot it changes color to indicate that it is the active target When you drop data columns onto a plot icon the plot icon is replaced by a plot of the dragged data W hen you drop data columns onto an existing plot the plot is redrawn using the new data columns To add data to a plot icon using drag and drop 1 Choose Tile Vertical from the Window menu so you can see your data and Graph sheet simultaneously 2 Select the data columns Use cTat click to select discontiguous columns 3 P
121. the specified colors Choose Special Colors to specify custom colors in a dialog Start M iddle End C olor Specify the boundary colors for the shading range A maximum of 64 colors are available for the color range choice Special Colors Specify custom colors in the standard Color dialog A maximum of 16 user defined colors are available T he title of this page varies by plot type It is available for Area Charts Bar Charts Contour Plots and Pie Charts You can specify standard fill attributes color pattern and background and border attributes color style and weight for each item Fill Color Specify the color to use for the filled region When you are choosing colors for your graph make sure that there is plenty of contrast in the colors you assign to each aspect of the graph If colors are too similar some parts of the graph will be difficult or impossible to see Fill Pattern Choose a pattern for the fill Pattern C olor Choose a color for your pattern Border Style Specify the line style to be used for borders around filled items or None to have no borders drawn Border C olor Specify the color for the border around a filled item Border Weight Specify a line weight for the border around a filled item PLOT PROPERTIES Area Charts Data to Plot To change the border and fill attributes for each item 1 Doubleclick on the plot to display the plot property dialog and choose the By Item page 2 Select the re
122. the start row Start Column Specify the location in the file of the column to begin reading from For example if you specify 5 S PLUS reads the columns beginning with column 5 and places them in the new data frame beginning at the Target Start Column Spreadsheet style letters e g A AB can be used to specify the start and end columns to import this works for any file type End Column Specify the location of the last column in the file to read Specify EN D if you wish to copy all columns after the starting column 121 CHAPTER 5 IMPORTING AND EXPORTING DATA Start Row Specify which row in the data file to begin reading from For example if you specify row 10 S PLUS reads the rows beginning with row 10 and places them in the new data frame beginning at row 1 End Row Specify the location of the last row to read Type EN D if you wish to copy all rows after the starting row Delimiter Specify all characters e g commas spaces periods used to separate elements in an ASCII file Commas spaces and tabs denoted by t are the default delimiters If you replace the default delimiters with another delimiter any column names or format strings you specify must be separated by the specified delimiter see the section Notes on Importing ASCII Delimited ASCII Files page 126 Carriage returns or line feeds are not allowed because they must terminate each row Note All controls on the Options page and
123. the top of the Contingency Table dialog Try changing this and opening the dialog The dialog will reflect the change T his change persists when S PLUS is exited and restarted 583 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE Status String This is currently empty Try entering text here do not forget to click Apply or OK and opening the dialog Property List This shows only the Property objects for the two tabbed pages SPropCrosstabsDataPage and SPropCrosstabsOptionsPage 10 more easily see these values right click in the edit field and select Zoom T he zoom box shown in figure 22 12 appears onm jE dit Bori Figure22 12 The zoom box howsthe Property Lig Using the O bject Browser open the property dialog for the first of these T his is shown in figure 22 13 Propart joci 1359 tare EA propery Lint a keai yu St NE Figure22 13 The Property dialog for the SPropC rostab dD ataPage Property object Argument List Use zoom if desired to view the assignments of Property object values to arguments of the function menuCrosstabs Notice in figure 22 11 that the return value is set to SPropSaveO bj T his has been done consistently throughout the user interface 584 DIALOGS Prompt List Since this is empty fields in the dialog will have their default prompts labels as specified in their corresponding property objects Default Value List Since this is empty fields in the dialog will have the default valu
124. to 0 5 Y Size Ratio to 1 0 and the Z Size Ratio to 1 0 Under Workbox Attributes set the Style to asolid line Click OK The plot should look similar to figure 9 7 Adjusting the angle of rotation 5 Click inside the workbox of the first panel Rotation handles will appear Drag one of the round handles counter clockwise The workbox will redraw in the new position The angle of rotation for all of the other panels will update to result in a complete rotation of TRELLIS EXAMPLES 360 degrees Figure9 7 Thegalaxy data rotated in 4 pands Vary the symbol color with velocity 6 Right click on any symbol in the plot to get the short cut menu Select Vary Symbols Set Vary Symbol Color to z Column Under Vary Colors set the First Color to Blue and the Last Color to Light Cyan Click on the Symbol tab Set the symbol height to 0 05 Click OK N ow with higher values of velocity a lighter colored symbol will be used Customizing the order of the panels We can use the customization options of the multipanel plot to skip the center panel and re order the contents of the panels so that they rotate around the center of the graph 7 Right click on a blank area within the graph to get the short cut menu Select M ultipanel Under Layout set of Columns to 3 and of Rows to 3 Under Continuous Conditioning set of Panels to 223 CHAPTER 9 WORKING WITH TRELLIS GRAPHICS Example 4 Using a User Defined Function to Add L
125. updated automatically when changes are made to the associated plot s properties To prevent the legend item from being deleted when the associated plot is deleted explicitly specify the number of legend items in the Legend dialog On the Font page you can specify standard font attributes for the legend item text Specifications made on the Font page apply to all of the legend item text ADDING LABELS FOR POINTS ADDING LABELS FOR POINTS If you have a 2D scatter plot on your graph you can automatically display labels determined by row names for selected points To label points 1 If you have more than one scatter plot on your Graph sheet select the scatter plot you want to use Click on the Label Points button in the middle of the top row of the Annotation Palette Click on a data point in your scatter plot A label will appear The label can be moved or edited as any other comment can To replace the label with a label for a different point click on another data point The first label will be removed and a new label will appear for the newly selected data point To add a label for another point shift click on another data point Another label will appear Identifying Points in a D ata View If you have a 2D scatter plot on your graph you can select rows in a data view in clicking on pointsin your scatter plot To select a row of data corresponding to a point on your graph 1 If you have more than one s
126. weighted least squares regression is calculated and the plot is created Plots2D 67 CHAPTER 3 TUTORIAL Changing Graph Features Axis and Labels Plot Properties 68 5 Click on the M aximize button in the upper right corner of the Graph sheet 6 Right click on a data point on the plot to access the plot s short cut menu Select Smooth Sort to try different levels of smoothing Enter values between 0 and 1 in the Span field and click on Apply or OK 7 Close the 2D plot palette S PLUS gives you unparalleled control over customizing every detail of your graph right down to the thickness of your tick marks You can control all individual line thicknesses symbol sizes fonts colors titles tick marks and axes labels Additionally you can create multiple lines of text for comments titles and tick labels Superscript and subscript options are conveniently located on a toolbar for quick access so editing any text or equation is easy Now we will change some of the features of the loess plot created in the previous section Go to the Windows menu and choose the Graph sheet GSn where n is the highest number you see to bring the window containing the 2D loess plot to the front of your screen M aximize the window 1 Right click on the y axis to get a shortcut menu Choose D isplay Scale 2 In the Display Scale page specify Color Lt Gray and Frame No ticks 3 Click on the Grids Ticks tab Under M ajor Ticks
127. will be the same as the range of your data 241 CHAPTER 10 FORMATTING A GRAPH 242 You can also specify the values for the beginning and end of the axis by entering anumber in this field For example if you specify an axis minimum of 1 and an axis maximum of 10 data points with the values of 1 and 10 will be positioned at the beginning and end of the axis respectively Major Tick Placement Interval For the Interval you can enter a number a range of numbers or choose from the following options Auto Data and Categorical Enter anumber to specify the exact number of intervals between ticks For example if you specify beginning and ending tick marks at 0 and 100 respectively see Tick Range and you specify 5 for the number of tick intervals then ticks will be placed at 0 20 40 60 80 and 100 If you specify a size of 20 with beginning and ending ticks at 0 and 100 respectively ticks will be placed at 0 20 40 60 80 and 100 as in the example above You can also specify a lower and upper bound for the number of ticks drawn separated with a colon for example 8 14 For example if you specify 8 as the lower bound and 14 as the upper bound S PLUS may produce 10 ticks depending on what is most suitable for your data Ranges are only valid if the Interval Type is Numbers If you choose Auto S PLUS will draw between 3 and 12 ticks depending on your data for most 2D plots T his is equivalent to specifying a range of
128. you can choose the and Apply OK Cancel or Apply button Buttons OK Choose the OK button or press ENTER to close the dialog box and carry out the action For example choosing the OK button in the Box Properties dialog will close the dialog box and accept the changes specified in the dialog Cancel Choose the Cancel button or press esc to close the dialog box and discard any of the changes you have made in the dialog Sometimes changes cannot be canceled for example when changes have made with Apply or when changes have been made outside of the dialog with the mouse Apply M ost of S PLUS s dialogs have an Apply button The Apply button acts much like an OK button except it does not close the dialog box You can specify changes in the dialog box and then choose the Apply button or press CTRL4ENTER to see your changes Keeping the dialog open so that you can make more changes without having to re select the dialog If no changes have been made to the dialog since it was last opened or applied the Apply button is dimmed 36 USING MENUS DIALOG BOXES AND TOOLBARS The Dialog Rollback Buttons Dialog Rollback The Dialog Rollback buttons let you restore a dialog to a prior state You can scroll back through each of the prior states until you find the set of values you want Then you can modify any of these values and choose Apply or OK to accept the entire current state of the dialog that is to change the corresponding object
129. your database vendor for specifics on this At this point a driver specific dialog should appear asking database and driver specific information required to connect to the database IMPORTING ODBC TABLES 32bit ODBC If you are running ODBC Administrator 2 0 you can create a User DSN D ata Source Name by selecting the Add button from the initial dialog Or to create a System DSN select that button and then the Add button on the subsequent dialog File D SN s are only available with O DBC 3 0 The type of DSN controls access to the Data Source you are creating Descriptions of the DSN types are available by selecting the Help button from the initial and System DSN dialogs Once you have selected either of the Add buttons select the driver that will export the data from the database to OD BC and select the O K button If the list of drivers is empty or does not contain a driver for your database or application you need to install the database or its ODBC driver Contact your database vendor for specifics on this At this point a driver specific dialog should appear asking database and driver specific information required to connect to the database Whether you are running O DBC 2 0 or 3 0 the new D ata Source should be visible the next time the Import ODBC or Export ODBC dialogs are selected from the file menu S PLUS has been tested with both ODBC 2 0 and ODBC 3 0 Table Name Once a data source has been selected t
130. 0 partial derivatives are not used The implied number of data points must be at least 2 but smaller than the number of input data points D raw Bars Choose whether to draw 3D bars If this option is not selected a 3D surface plot is drawn X and Y Bar Width Specify the width of the X and Y bars asa fraction of the grid interval For example an X bar width of 0 75 specifies that the bars should occupy 3 4 of each grid interval in the X dimension with a gap between bars of 1 4 of the interval 3D bar charts are constructed just like 3D surface plots except that they have bars drawn straight down from the grids As a result there are no By Bar specifications available for 3D bar charts You can create a 3D bar chart by selecting data and choosing the 3D Bar Chart button on the Plots 3D palette T his automatically specifies the D raw Bars and X and Y Bar Width options on the 3D Bars page of the property dialog If these are not specified a surface plot is drawn 3D spline plots are surface plots that are smoothed using cubic spline interpolation You can create a 3D Spline by selecting data and choosing the Spline button on the Plots 3D palette This automatically specifies the desired number of X and Y Data Grids to be displayed on the surface A spline plot is created if the 0 utput Grids exceeds the number of D ata Grids 335 CHAPTER 12 EDITING PLOT PROPERTIES 336 EXPORTING GRAPHS PRINTING AND SENDING MAIL Selectin
131. 25 pre sorting data 2D 321 3D 330 principal components 500 projecting onto existing 3D graph 202 using the menus 202 projecting plots 201 projecting plots onto a plane 203 proportions test 402 Q quantile quantile plot 418 quantiles 355 R random effects analysis of variance 450 random number generation 358 random number seed 354 random samples 353 range shading 302 redrawing automatic 604 manual 604 regression plots 313 regression plots 3D base line level 330 Report window 563 operations 563 residual fit spread plot 418 resizing fonts with graph 601 symbols with graph 601 robust linear regression 428 rotating comment plot text 310 row selection 91 Row s or if 301 rows deleting 105 inserting 103 613 INDEX rugplot 418 449 Run Script button 549 S sample size random 358 sampling probabilities 353 Save As field 342 Save In fidd 342 saving to different formats exporting See saving defaults columns 115 data sheets 111 scatter plots 319 data specifications 319 lines 319 pre sorting data 321 symbol color 320 text as symbol 320 scatter plots 3D colors 331 data specifications 330 plot options 330 pre sorting data 330 regression 331 symbols 330 vary symbols 331 Script 555 script files 548 script window 547 Script windows 545 context menus 555 dragging objects into 560 Expand I nplace 557 font 557 output pane 547 program pane 547 614 scripts editing 548
132. 30 Line amp Scatter Plots 3D 330 Surface Plots 332 Chapter 13 Exporting graphs Printing and Sending Mail 337 Printing Sheets and Scripts 337 Sending Electronic M ail 338 Exporting Graph Sheets to D ifferent File Formats 339 Chapter 14 Using the Statistics Menus and Dialogs 341 Introduction to Statistics M enus and Dialogs 341 Dialog Fields 342 Plotting from the Statistics D ialogs 343 Saving Results From an Analysis 343 CONTENTS S PLU S Functions Called by Statistics D ialogs 344 M odifying the Statistics D ialogs 345 Chapter 15 Creating and Manipulating Data 347 Introduction 347 N ew Data O bject 348 Tabulate 349 Merge T wo D ata Frames 351 Random Sample Generation 353 Density Cumulative Probability or Quantile 355 Random Number Generation 358 Chapter 16 Summarizing and Exploring Data 361 Summary Statistics 362 Contingency Table 365 Correlations and C ovariances 368 Local Regression LO ESS Smoothing 370 Supersmoother 372 Kernel Smoother 374 Spline Smoother 377 Chapter 17 Comparing Samples 379 One Samplet T est 380 One Sample Wilcoxon T est 382 One Sample Kolmogorov Smirnov T est 384 One Sample C hi Square Goodness of F it T est 386 Two Sample t T est 388 Two Sample Wilcoxon T est 390 Two Sample K olmogorov Smirnov T est 392 One Way Analysis of Variance 394 Kruskal Wallis Rank Sum T est 396 Friedman Rank Sum Test 398 Exact Binomial T est 400 Proportions T est 402 Fisher s Exact T est 404 M cNemar s Chi Square
133. 316 cell selection 90 Chi Square test 367 chi square tet 386 606 color range 302 surface plots 334 Color Schemes dialog 603 Color Styles dialog 602 colors line 301 symbols 301 column data type changing 114 column display precision 115 column list 95 column names 113 editing 113 editing interactively 114 column numbers 94 column selection 91 column width 112 113 columns adjusting width to widest call 113 changing width by dragging 112 with the toolbar 113 deleting 104 description editing 113 display format 114 formatting 112 moving 101 overview 94 precision 115 saving defaults 115 selecting from lists 95 specifying multiple 197 using for pie chart labels 325 combining existing graphs 192 multiple 2D plotsin 3D space 201 combining 2D and 3D plots 203 Commands window overview 561 Commands window example 81 Commands window settings 600 INDEX comment plots 309 box margin 310 font styles 310 justification 310 offsetting text 310 plot options 310 rotating comment text 310 text display format 310 text display precision 310 units 309 compare models 485 comparisons 48 1 conditioning columns 601 conditioning mode 601 Conditioning M ode button 209 confidence interval type 508 519 confidence level 380 389 error bar plots 317 confidence limits 425 connect type 301 contour levels 313 contour lines 313 contour plot labels colors 312 display format 312 frequency 312 precision 312 contour plots 311
134. 3D graph from different viewing angles X Angle 10 X Angle 100 Figure 10 6 3D surface plot viewed from different angles To rotate a 3D graph in multiple panels 1 Createa3D plot 2 Select the graph area and click one of the Panel Rotation buttons on the plot palette Or 256 DISPLAYING 3D MULTIPANEL GRAPHS w Doubleclick on the graph area to open the Graph Properties dialog On the Multipanel page check the Rotate 3D Axes checkbox and specify the number of panels You can specify the amount of rotation to be used across all of the panels in the Rotate Panel O ptions fields on the 3D Workbox page Displaying 3D You can use multiple panels to show your 3D graph sliced along one axes Sliced Graphs ZO Oral Zee tonal Z Sno NS Zi to 01 Figure 10 7 The previous figure diced along thez axis 257 CHAPTER 10 FORMATTING A GRAPH To slice a 3D graph in multiple panels 1 Createa3D plot 2 Select the graph area and click one of the Panel Slicing buttons on the plot palette Or w Doubleclick on the graph area to open the Graph Properties dialog On the Multipanel page specify one of the columns of data to be used as the conditioning variable in the Column List field 258 FORMATTING POLAR AXES FORMATTING POLAR AXES Display Page Ranges Ticks Page To select the axes w Click the horizontal axis To format the axes 1 Doubleclick the horizontal axis or select the axis and choose Sel
135. 409 CHAPTER 17 COMPARING SAMPLES PEARSON S CHI SQUARE TEST 410 Data Results This dialog performs Pearson s chi square test on a two dimensional contingency table See the on line help for chisq test for details on the assumptions of the test and the null hypothesis To perform Pearson s chi square test Choose Statistics gt Compare Samples gt Counts and Proportions gt Chi Square Test from the main menu T he dialog shown below appears Pearoon s Chi est BEA pem F Appi reiss Conbnuty Consi Cisis Preme Pemb Ciscaiinshon visraiic Osa Save Ad z F Priel Faas Chea ifinstion veraio Teo Data Frame Select a data frame Classification Variable One Select the factor column that contains the first classification variable It must have at least two levels Classification Variable Two Select the factor column that contains the second classification variable T his variable must have at least two levels Apply Yates Continuity Correction Check this if the results to apply Yates correction for continuity See the on line help for chisq test for an algebraic definition of the continuity correction Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten The saved object will have classht est PEARSON S CHI SQUARE TEST This must be a valid S PLUS object name any combination of alp
136. 590 M anipulating M enu Items 590 Saving and O pening M enus 591 Example Customizing the C ontext M enu 591 Using the Class nfo O bject 595 Properties of the C lassi nfo O bject 595 Creating and M odifying a ClassInfo O bject 596 In S PLUS it is easy to modify the appearance of the dialogs found in the user interface T he user can also create customized dialogs and invoke them with toolbar buttons and menu items Similarly menus toolbars and palettes can be created and modified by the user This chapter describes in detail how to use the S PLUS user interface to create and modify the dialogs menus toolbars and palettes which make up the interface 567 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE TOOLBARS AND PALETTES Creating Toolbars In S PLUS toolbars and palettes represent the same type of object When a toolbar is dragged into the client area below the other toolbars it is displayed there as a palette W hen a palette is dragged to the non client area close to a toolbar or menu bar it docks there as a toolbar Toolbars are represented in the Object Browser as Toolbar objects T hese contain ToolbarButton objects which represent their buttons This section shows how to work with toolbars through the user interface W hile it isnot hard to create or modify toolbars through the user interface as shown in this section it is sometimes easier to work with toolbars using the S PLUS programming language As you work
137. 7 O a Glabron 4 O A Peatland O A Velvet o a O Manchuria Pa a Svansota _O a 20 40 60 yield Figure 9 6 Trellis graphics were used to pinpoint an error in Barley yield data 221 CHAPTER 9 WORKING WITH TRELLIS GRAPHICS Example 3 3D Plots of Galaxy Velocity 222 T he data frame galaxy contains measurements on the velocity of NGC 7531 a spiral galaxy in the Southern H emisphere T here is substantial variation in the velocity at different locations measured by east west and south north positions which are measured in arc seconds In this example we will examine how the velocity measurements vary over the measurement region Creating a 3D scatter plot and surface plot 1 Select the galaxy data frame in the left pane of the O bject Browser Click on east west then CTRL click on north south and velocity to select the three variables Click on the 3D plot palette button on the Standard toolbar to open the palette Click on the 3D scatter plot icon to create a 3D scatter plot of the data 2 Maximizethe Graph sheet Rotating the graph in panels 3 Click anywhere inside the graph to select it Then click on the 4 Panel Rotation button on the 3D plot palette The graph will be redrawn 4 timesin different panels at different rotation angles Adjusting the shape of the workbox 4 Right click on a blank area within the graph to get the short cut menu Select 3D Workbox Under Workbox Shape set X Size Ratio
138. BACKSPACE and DELETE keys to erase typing errors and the mouse or RIGHT and Lert direction keys to move around in the entry To cancel an entry when typing press Esc N ote that for factor data there is the option of selecting the factor level from a drop down list if it has been entered or specified previously 3 Press ENTER UP Or DOWN to enter the data in the cell If you press a direction key the cell cursor moves in the appropriate direction The cert direction key accepts and enters the value in the cell if pressed when the cursor is at the far left of the entry and the RIGHT direction key accepts and enters the value in the cell if pressed when the cursor is at the far right of the entry If you begin typing in a cell that already contains a value the old value is overwritten ENTERING DATA Entering Data from Other Sources Editing Data There are other methods of entering data into S PLUS besides typing it in from the keyboard T he easiest way to enter datais to import it from another source You can import Excel Lotus SAS ASCII and dBase files among others See the chapter Importing and Exporting data for detailed information on importing data Data can also be entered by using options in the Data menu You can use Copy and Move to get data from other columns into the current Data window You can also write formulas in the Commands window that enter data into specified columns For example you could write a
139. CENSE AGREEMENT AND LIMITED WARRANTY Il License Grant lll Ownership of the Software IV Warranty and Remedies 618 A MathSoft grants to Licensee a personal non exclusive and non transferable license to make one copy of the Software for use solely for the purposes described in this Agreement and make one copy of the Software for archival purposes B If a single user license has been purchased this license is limited to loading and using the Software on a single physical workstation C If a network or server based license has been purchased the number of simultaneous users of the Software must not exceed the number of software licenses purchased by the Licensee D The Licensee will not transfer the Software to any other party except with written authorization from MathSoft A It is expressly understood and agreed that all right title and interest in and to the Software and any other material furnished to Licensee under this Agreement vest solely and exclusively in MathSoft and Licensee shall neither derive nor assert any title or interest in or to such items except for the rights and licenses granted under this Agreement B Under this Agreement Licensee does not receive any rights to patents copyrights trade secrets trademarks or any other rights or licenses to the Software beyond those expressly granted in this Agreement A MathSoft warrants that the physical software media and the documentation will be free fr
140. Choose Stagger to have your tick labels shifted up and down alternately zigzagged along the axis You need to specify the depth of staggering in the Number field For example if you specify 2 your tick labels are staggered two levels Auto will choose the amount of staggering required to prevent overlapping Choose Truncate to have your tick labels cropped abbreviated to avoid overlapping Choose None to turn off all placement options Labels may overwrite each other Number Specify the number of labels you want to skip or the depth of staggering Choose Auto to allow S PLUS to determine skip or stagger details to avoid overlap Adjust Numeric Labels Enables or Disables shrinking of labels If shrinking the labels causes other labels on that axis to overlap S PLUS will shift or skip the labels according to your specifications see O velap H andling in this section Shrink Factor The shrink factor specifies the amount tick labels can be reduced to prevent overlapping A value of 1 representing 100 specifies that labels are to be drawn at their specified height even if they overrun one another T he default value of 0 8 or 80 allows S PLUS to shrink the labels up to 20 of their specified height so they will not overlap i e labels will be drawn at least to 80 of the specified height and will be drawn larger if possible If you specify 0 the labels will be reduced until they no longer 247 CHAPTER 10 FORMATTING A GRAPH
141. Continuity Correction 403 Trademarks e S PLUS is a registered trademark and StatServer S HNTERFACE S SPATIALSTATS S GISLINK SHDOxX S WAVELETS and AXUM are trademarks of M athSoft Inc S and New S are trademarks of Lucent Technologies Inc Intel is a registered trademark and 486 SX and Pentium are trademarks of Intel Corporation M icrosoft Windows MS DOS and Excel are registered trademarks and Windows NT isa trademark of M icrosoft Corporation SAS is atrademark of the SAS Institute Inc SPSS is aregistered trademark of SPSS Inc All other trademarks are acknowledged License Agreement and Limited Warranty I Notice A IM PORTANT Before starting the installation process you will be asked to accept the terms of this Agreement Read this Agreement carefully before completing the installation process BY COMPLETING THE INSTALLATION PROCESS OR BY HAVING AN AGENT SUCH ASA COMPUTER TECHNICIAN DO SO FOR YOU YOU ARE AGREEING TO BE BOUND BY THE TERMSOF THIS AGREEMENT This Agreement is a legal contract which specifies the terms of the license and limited warranty between you Licensee and Mathsoft Inc MathSoft for the MathSoft S PLUS software the Software and the associated documentation software and documentation collectively referred to as the Licensed Works If you do not agree to the terms of this Agreement promptly return all copies of the Licensed Works to MathSoft 617 LI
142. Do oo a Mo LK oH DOD oV OVT OY GB GOV GS GV GS PRODI RB RODD RID Date Figure8 5 Multiple plots on a sngle graph Adding a Plot Using a Plot Button Each plot on a graph represents one or more columns of data from a data sheet T he plots can all be of the same plot type for example line plots or be a combination of different plot types for example line scatter and bar plots Combined plots must have the same type of axes For example both a line plot and a bar chart have XY axes and can be combined on one graph H owever a box plot and surface plot cannot be combined on the same graph because they have different types of axes A 2D graph and a 3D graph can both be placed on the same Graph sheet but will not be on the same graph See the section Adding a Graph to an Existing Graph Sheet page 190 for details In addition 2D plots can be projected onto 3D planes See section Projecting a 2D Plot onto a 3D Plane page 201 later in this chapter You can easily add plots to an existing graph by selecting the graph selecting the data and sHiFt clicking on plot buttons 183 CHAPTER 8 CREATING A GRAPH Adding a Plot Using Drag and Drop 184 To add a plot using a plot button 1 Select the graph to which you want to add the plot 2 Open aD ata window containing the data to plot or select the data Set in the O bject Browser so that the columns appear in the right pane 3 From the Window menu choose Tile V
143. E SURVIVAL AND TIME SERIES MODELS Model Optom Fetus ontrasis agata 5 waishi Conesa a Options page Contrasts Assign Contrast Choose contrasts for the factors by default the Helmert contrasts are assigned to unordered factors and polynomial contrasts are assigned to ordered factors to Variable s Select one or more variables to which the selected contrast in Assign Contrast will be assigned Contrasts This field displays the selection and assignment chosen in Assign Contrast and to Variable s 490 MULTIVARIATE ANALYSIS OF VARIANCE Results page Printed Results Saved Results kemena Anahit of VAN aN EE Made Optora Pera ult Pramis Pena te Saeed Aes F ihipa Ea in j ARO Tada D Aieiaia Tired rey witty pnei VD Assetuels P Ensari Comines l Ewteiaied E Coul tor foral Facio Short Output Check this to print the call to the S PLUS function maov the sums of squares and degrees of freedom for factors and residuals and the residual standard error ANOVA Table Check this to print an AN OVA table Testing with Select the type of test used in the ANOVA table Estimated Coefficients Check this to print the estimated coefficients T here are K 1 such coefficients for each K level factor Estimated K Coef for K level Factor Check this to print K coefficients for each K level factor Fitted Values Check this to save the fitted values Residuals Check this to save the re
144. ELETING CELLS ROWS AND COLUMNS Inserting Columns and Rows 102 In S PLUS you can insert columns and rows between existing columns and rows in your data sets When you insert columns existing columns are shifted to the right to make room for the new column When you insert rows existing rows are shifted down to make room for the new rows W hen you insert blocks existing rows and columns are shifted down and to the right to make room for the new cells To insert a column 1 Select the column you want to have shifted to the right to make room for the new column 2 Choose the data type for the new column from the drop down list next to the Insert Column button see figure 4 5 3 Click the Insert Column button You can also specify formatting information when inserting a new column or row by using the Insert dialogs T he Insert dialogs are accessed through the Insert pull down menu in the main menu bar Specific to inserting columns the Fill Expression field allows you to fill the column with some form of initial values For example to fill all the column with zero s simply enter o in this field To enter random numbers from a standard normal distribution in a column of length 100 you could enter rnorm 100 If the column is of length 200 then this sequence of random numbers would be repeated But note that if you wish this sort of repetition the expression used must divide exactly into the column length that is rno
145. ELS Model Page Select Model This dialog compares models using analysis of variance type methods It calls theanova function which will call the appropriate method function for the class of models being compared For example anova m for comparing m linear model objects M ode objects being compared must all have the same class and the same response variable The comparisons only make sense if the models are nested one model is a subset of the other To compare models Choose Statistics gt Compare Models from the main menu T he dialog shown below appears Compa Medel Miteleosd Aaen les ma Fa z Radts Save Ag are FF Pint espa D ea are of ome Hp Model Objects Select the model objects for comparison An arbitrary number of models can be selected Name String Match Enter a pattern used to restrict the list shown in the Model O bjects drop down list The symbol matches any character For example to view all objects that begin with last enter last Use to denote a list of character options For example modell model2 and model3 match model 123 but model4 does not 485 CHAPTER 18 FITTING STATISTICAL MODELS 486 Results Model Class Select the class of models you want to do comparisons with All models being compared must be from the same class Test Statistic Select the test statistic to use for the model comparison Note that the availa
146. Each case is selected with a probability equal to prop The second samp_fixed sample size total observations selects a random sample of fixed size The first case is drawn with a probability of sample_size total_ observations and the succeeding ith case is drawn with a probability of sample size hits total_ observations i Finally a third function samp_syst n performs a systematic sample of every nth case after a random start Expressions are evaluated from left to right so you can sample from a subset of your cases by subsetting them first and then sampling For instance to take arandom half of high school graduates use where schooling gt 12 amp samp_rand 5 125 CHAPTER 5 IMPORTING AND EXPORTING DATA NOTES ON IMPORTING FILES Notes on Importing ASCII Delimited ASCII Files 126 When importing ASCII files you have the option of specifying column names and data types for imported columns This can be useful if you want to name columns or if you wish to skip over one or more columns when importing If columns are specified one by one here the END Column specification is ignored In the Import Data dialog the Options page has the following ASCII options Column Names Enter alist of column names for the data columns to import separated by any of the delimiters specified in the Delimiters field Specify one column name for each imported column eg Apples O ranges Pears You can use an aster
147. Enter the top coordinate of the toolbar in pixels Toolbar Left Enter the left coordinate of the toolbar in pixels Button Rows Enter the number of rows of buttons in the toolbar Using the Property Dialog Right click on the Toolbar object in the O bject Browser New Toolbar Select this to open a new toolbar New Button Select this to add anew button to the toolbar Hide Select this to hide the toolbar Delete Select this to delete the toolbar Open Select this to open a toolbar that has been saved in an external file Save Select this to save a toolbar to its external file when one exists Save As Select this to save a toolbar to an external file Unload Select this to unload a toolbar from memory The toolbar is no longer available for display To reload a built in toolbar restart S PLUS To reload a toolbar that has been saved to an external file open that file Restore D efault Toolbar Select this to restore a built in toolbar to its default state after it has been modified Properties Select this to display the property dialog for the Toolbar object Buttons Select this to display a dialog used for displaying or hiding different buttons on the toolbar 573 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE Displaying Toolbars Refresh Icons Select this to refresh the icon images on the toolbar buttons after they may have been modified Save Toolbar O bject as default Save a modified version of a toolbar as the default for
148. ICAL MODELS MULTIPLE COMPARISONS T his dialog calculates simultaneous or non simultaneous confidence intervals or bounds for any number of estimable linear combinations of the parameters of a fixed effects linear model It calls the mul ti comp function and its print and plot methods To perform multiple comparisons Choose Statistics gt Multiple Comparisons from the main menu The dialog shown below appears Multiple Comparisons Confidence Intervals OE E3 Model Options m Model Selection gt p Error Specification Model Object oOo y Confidence Level 0 95 Name String Match ts Bounds Jupper and lower Variable Selection Error Type fiamily wise gt Compare Levels Of a Results Compare To Les TE Savane flast multicomp r Method IV Print Results Comparisons mca z IV Plot Intervals Method best fast al o Cancel Apply current Help Model Page Model Selection Model Object Select the model object on which to perform multiple comparisons Name String Match Enter a pattern used to restrict the list shown in the M odel O bject drop down list The symbol matches any character For example to view all objects that begin with last enter last Use to denote a list of character options For example modell model2 and model3 match model 123 but model4 does not 480 MULTIPLE COMPARISONS
149. K To move the graph area w Select the graph area and drag it to a new location or 233 CHAPTER 10 FORMATTING A GRAPH w Doubleclick inside the graph area to display the Graph Properties dialog Click on the Postion Size page For the Graph Position specify the H orizontal and Vertical position and chooseO K To format the graph area 1 Doubleclick inside the graph area to display the Graph Properties dialog 2 Click on the Fill Border page M ake the desired formatting changes to the graph area 3 ChooseOK Formatting the You can select the plot area to change its size or formatting In a2D graph the Plot Area plot area is bounded by the axes Figure10 2 The plot area gray is sdected To select the plot area w Click anywhere inside the region bounded by the axes When selected the plot area has knobs on all sides and all four corners To change the size of the plot area w Select the plot area and drag the corner resize knobs to the desired size or 234 FORMATTING A GRAPH Position Size Page w doubleclick inside the graph to display the Graph Properties dialog Click on the Position Size page For Plot Display Size specify the H eight and W idth and choose OK or w select the plot area right click and select Position Size For Plot Display Size specify the H eight and Width and choose OK To move the plot area w Select the plot area and drag it to a new location or w double click in
150. LUS to suit your working style This chapter gives an overview of the menus windows and toolbars that are the backbone of the product Be Edi er ose Format Gate partits Graph Opin adie Helo lee CES Aa Pr ae Linkerccepea IT redistion 7 temp rstoza 1 05 P e itial stemdard imati 0 5102 an LI random Bulcipia R Squered 0 OIT F arariari 16 035 on 3 ard 107 the p valus is u Heien MUH So bey lba Figure2 1 S PLUS in action notetheGraph het top left with the3D plot palete the O bject Browser top right a Data window beow left and Report window below right The three variables columns of the data set ethanol have ben sdected in the D ata window and are high lighted and the resulting 3D scatter plot isshown in the Graph sheet GS1 28 USING MENUS DIALOG BOXES AND TOOLBARS USING MENUS DIALOG BOXES AND TOOLBARS Using the Mouse Mouse Pointers S PLUS menus dialogs and toolbars contain all the options you need to manipulate data create stunning graphs and write S PLUS scripts You can use your mouse or your keyboard to access S PLUS s menus Dialogs can be accessed be selecting menu options or by double clicking on objects M ouse keyboard and window terms used throughout this User s Guide are defined below Throughout this U sr sGuide the following conventions are used to describe mouse operations Pointing M oving the mouse to position the pointer over an object When pointing and
151. MA model Choose Statistics gt Time Series gt ARIMA Models from the main menu The Model Page dialog shown below appears Modai Fite Optra Diagoaica Forcent Time Senes Dais imis Pesse Ophonal Daia Fiama l Enter bite Fermamebs vales Tms Sere ayi ARPA Model Onia aT Astoregreese ip fi l otaran Citerence jd F j T Add a Tine Sarii Coser Mring Avice ieh f SAMA Model Pennie ai Sorsntakiy E Hone 15 ESETT fission F Dianai i F Previous P hanik 12 r ohai Time Series Data Data Frame 528 Time Series Select or enter the name of a data frame having time series as columns ARIMA MODELING Select the column containing the time series to be modeled gt Tip You can enter the name of a univariate time series object directly in Time Series ARIMA Model Autoregressive p Order Enter an integer giving the order of the autoregressive operator Difference d Enter an integer giving the number of differences Moving Average q Enter an integer giving the order of the moving average ARIMA Model Seasonality Periodicity Select the period of the seasonal operator Period Select the period of the seasonal operator of the ARIM A model T his field is needed if Seasonality is O ther Initial Parameters Enter Initial Parameter Values Optional Check thisto enable the input of initial AR coefficients and M A coefficients AR coefficients Enter the vector of initia
152. MATTING 2D AXES Grids Ticks Page Axis Breaks Page column of numbers for variable interval widths Choose Factor 10 to place ticks at intervals of powers of 10 for example 1 10 100 1000 Choose Factor e to place ticks at intervals of powers of e Tick Range Specify the positions of the major First Tick and Last Tick Specify numbers in axis units to choose particular values for the first and or last major tick marks Select Auto to place the first and last ticks either outside of or on the first and last data point Select Axis to place first and or last tick marks at the axis endpoints Select D ata to place first and last ticks at the first and last data points Select InsideD ata or O utsideD ata to place the first and or last tick mark inside or outside the data range respectively A grid is a set of lines that extend from the tick marks across the graph In the Axis dialog the Grid T icks page has the following options Major and Minor Grid Attributes Specify the style color and weight for the major and minor grids To turn the grids on or off use the State option Tick Length and Tick Weight Specify the length and weight of the tick marks The length is measured in relative inches cm depending on which you specified Tick Position Specify the orientation of the tick marks Specify whether to have ticks drawn from the axis In from the axis O ut to have ticks C ross the axis or not drawn at all O ff Panel T
153. MATTING A GRAPH Fill Border Page Plot Summary Page Fill Border Page Multipanel Page 236 Aspect Ratio This group controls the size of the axes area within the designated plot area Choose Auto to have the aspect ratio calculated automatically Choose Fill Plot Area to use the full extent of the plot area for the axes Choose Banking to have the aspect ratio adjusted to normalize the average slope of all line segments of all appropriate plots in all panels to the value specified in the Ratio Value field The Ratio Value is interpreted in degrees and may be specified as Auto Choose Proportional U nits to force an aspect ratio that causes the units for both axes to be in proportion to one another by a ratio specified in the Ratio Value field For example a value of 1 will cause the units of both axes to be interpreted as equal Choose Ray From Origin to cause the aspect ratio to be adjusted such that a ray from the axes origin through the point specified by the X Value and Y Value fields will pass through the point defined by the axes maximums C hoose Ray From 0 0 to cause the aspect ratio to be adjusted such that a ray from the point 0 0 through the point specified by the X Value and Y Value fields will pass through the point defined by the axes maximums Plots that are used in the automatic aspect ratio calculations contain a U se for Aspect Ratio check box on their D ata To Plot pages T hese plot types include Line Scatter Smoo
154. MathSoft S PLUS User s Guide Version 4 5 M arch 1998 D ata Analysis Products Division M athSoft Inc Seattle Washington Proprietary Notice Copyright Notice Acknowledgments M athSoft Inc owns both this software program and its documentation Both the program and documentation are copyrighted with all rights reserved by M athSoft The correct bibliographical reference for this document is as follows S PLUS Users Guide Data Analysis Products Division M athSoft Seattle WA Printed in the United States Copyright 1997 1998 M athSoft Inc All Rights Reserved S PLUS would not exist without the pioneering research of the Bell Labs S team at AT amp T now Lucent Technologies Richard A Becker John M Chambers Allan R Wilks William S Cleveland and colleagues This release of S PLUS includes specific work from a number of scientists The cluster library was written by Mia Hubert Peter Rousseeuw and Anja Struyf U niversity of Antwerp Updates to functions provided to this and earlier releases of S PLUS were provided by Brian Ripley Oxford University and Terry Therneau M ayo Clinic Rochester USER S GUIDE CONTENTS OVERVIEW Introduction Chapter 1 Welcome to S PLus Chapter 2 Working with the Graphical User Interface Chapter 3 Tutorial Data Management Chapter 4 Using Data Windows Chapter 5 Importing and Exporting Data Chapter 6 Using The Object Browser Chapter 7 Exchanging
155. O bject Graph2D Graph3D GraphSheetPage Graph2D 2D plots 2D axes panel strip LinePlot LinearCF Plot AreaPlot BarPlot BoxPlot CommentPlot C ontourPlot ErrorBarPlot GraphM atrix Histogram PiePlot QQ Plot Axis2dX Axis2dY PanelStrip Graph3D 3D plots 3D axis panel strip ContourPlot Line3D Plot SurfacePlot Axis2dX Axis2dY 2D plots and 3D plots GraphSheetPage Folder SearchPath M enultem Toolbar axis label and title Trellis panels page itens folder items database menu items toolbar buttons Axis3D PanelStrip Axis2D LabelX Axis2D LabelY AxisT itle GraphSheetPagel tem Folderltem D ataBase M enulten ToolbarButton An object of class Menultem becomes a parent for other M enultems by setting its Type property to M enu 145 CHAPTER 6 USING THE OBJECT BROWSER The Default Object Browser The Examples Object Browser To Find S PLus Objects 146 TheG raphSheetPagel tem and Folderltem objects are shortcuts to objects A default O bject Browser is provided and is opened by selecting the toolbar button of the Standard toolbar The tool tips can aid in finding the appropriate toolbar button Generally the default O bject Browser is created when S PLUS is started but can be disabled in the Startup tab of the G eneral Settings dialog that can be activated by the O ptions G eneral Settings menu item T he default browser configuration file is named D e
156. O bject Browser Show Dialog On Run This is relevant when Action is set to FUNCTION Check this to cause the dialog associated to the function to open when the item is selected TOOLBARS AND PALETTES Always Use D efaults This is relevant when Action is set to FUNCTION Check this to force the use of the default values when the function executes This can also be specified through the context menu S PLUS makes a distinction between the default argument values for a function as defined in the function s dialog via the Functi onl nfo object and as defined by the function itself Always Use D efaults refers to the dialog defaults Table 22 1 below summarizes how Show Dialog On Run and Always Use Defaults work together In it function refers to the S PLUS Table 22 1 Summary of Show D ialog On Run and Always U D efaults Show Dialog On Always U se ne R n D efaults Action when the menu item is selected checked checked The dialog always opens in its default state when the menu item is selected Changes are accepted but do not persist as dialog defaults checked unchecked The dialog always opens when the menu item is selected Changes are accepted and persist as dialog defaults unchecked checked The dialog does not appear and the function executes using the current dialog defaults unchecked unchecked The dialog will appear once either when the menu item is selected or when Customize is selected from the menu
157. Output Grids Specify the desired number of X and Y grids The finer the mesh you specify that is the greater the number of specified output points the more difficult it may be for the algorithm to obtain reasonable estimates of the contour levels at the grid points H owever the coarser the grid you specify the cruder will be the approximation to the contour T hus some care should be taken in specifying the number of output points X and Y Minimum and Maximum Specify the range of X and Y for which you want the contour drawn When Auto is specified the minimum and maximum values in the source X and Y columns are used Grid D ata Specify N o if your data columns are already gridded Specify Yes if your data columns are irregular If you choose Auto S PLUS will try to determine if the data are gridded or irregular If the data are irregular S PLUS will grid the data automatically If your data are gridded and in the long format it is recommended that you specify N o instead of Auto in this field Algorithm Bivariate uses the interp function to compute a bivariate interpolation for irregular grids T he triangulation scheme used by interp works well if x and y have similar scales but will appear stretched if they have very different scales The spreads of x and y must be within four orders of magnitude of each other for interp to work Internal uses a cell search algorithm internal to the graphics It essentially searches for the three points
158. PLUS U s s Online Manuals Guide are all available online as well as in print To view a manual online select O nline M anuals from the S PLUS Help menu and choose the desired title Notes on Online versions of the Guides The Online manuals are viewed using Acrobat Reader which can be installed as an option during the installation process While using Acrobat Reader it is generally useful to turn on bookmarks under the View entry of the menu bar rather than rely on the contents at the start of the guides Bookmarks are always visible and can be expanded to include section headings or collapsed to show just chapter titles Online Demo TheS PLus Online D emos help users of all levels familiarize themselves with the new features of S PLUS Take a look at the user interface learn more about common S PLUS tasks or show a colleague the various capabilities of S PLUS Guided Tours of S PLus Usa s Guide contains a tutorial and many chapters have examples S PLus using S PLUS T hese examples extend the techniques illustrated in the online demos 19 CHAPTER 1 WELCOME TO SPLUS Add On Modules StatLib S News Training Courses 20 Add on modules that offer analytical functionality beyond that of the base S PLUS product include S D OX helps in designing and analyzing industrial experiments especially fractional factorial experiments response surface experiments and robust design experiments S GARCH prov
159. PTER 20 BUILDING FORMULAS Survival Response Add Term s Remove Term Examples 542 The fields necessary to define the survival response will vary depending on the type of censoring for the problem at hand The Time 2 and Origin for H azard fields will only be enabled for a counting process Choose a time variable from the variable list Survival Times or Start Time and click on the Time 1 button Similarly select a variable representing the Censoring Indicator event If appropriate select a variable for Time 2 Ending Time After all required variables have been selected click on the Add Response button to place the survival response in the formula Choose one or more variables from the Choose Variable s drop down list and then click M ain Effects or M ain Hnteract button T he appropriate main effect and or interaction terms will be added to the formula Click on the Cluster Strata or Offset button to define clustering stratifying or offset variables respectively T he appropriate term will be added to the formula As predictors are added to the formula they are listed as options in the Term list box To delete a given term select a term from the list then click on the Remove button Formula The wide string Formula box shows the formula you are building You may edit the box directly or right click on the box and then click on zoom for further editing Example 1 A simple case using the Ovarian data frame Th
160. Page You can import graphics using the Insert Picture option see below or from a script using the gui Create command See the S PLUS Programme s Guide for more information on gui Create To import a graphic image file 1 With a Graph sheet in focus choose Picture from the Insert menu 2 In the Filename box type or select the name of the file you want to import 3 In the Files of type box select the type of the file being imported Choose O pen to import the file To change formatting for the inserted image w Right click on the image to display its shortcut menu or double click on the object or select the image and choose Selected Insert Picture from the Format menu M ake the desired changes and choose OK You can specify standard fill attributes fill color fill pattern and pattern color and standard line attributes style color and weight for the graph area In the Picture dialog the Position page has the following options X Y Specify the location for the lower left hand corner of the image Width Height Specify the width and height for the imported image in document units e g inches If either value is set to zero S PLUS will use the default width and height for the object itself e g the default size of a bitmap Use Axes Units C hoose to position the picture relative to an axes pair on the graph If you choose this option your x and y data values are interpreted just like x and y data points on the g
161. R 17 COMPARING SAMPLES ONE WAY ANALYSIS OF VARIANCE 394 Data Results This dialog generates a simple analysis of variance table when there is a grouping variable available for the data and this grouping variable defines separate k samples of data No interactions are assumed among the main effects that is the k samples are considered independent To perform a one way analysis of variance Choose Statistics gt Compare Samples gt k Samples gt One way ANOVA from the main menu T he dialog shown below appears Une wen Anskes of wananipa Data Fria Daia Frene See AT Fiasgan ariana F Pmi Gesets Grouping engin Data Frame Select a data frame Response Variable Select the column that contains the response variable T his must be numeric Grouping Variable Select the column that indicates group membership for the response above T his is usually a factor variable with k levels Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten The saved object will have class a ov This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names Print Results ONE WAY ANALYSIS OF VARIANCE Check th
162. Report window toolbar see chapter 21 Working with Script Commands and Report windows Turn on context help Commands Window Toolbar Table 2 9 T he Commands window toolbar see chapter 21 Working with Script Commands and Report windows Create object oriented editable graphs 51 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE Plots2D Palette Scatter Line Isolated Points Bubble Loess Linear Fit QQ Normal Plot w Line Histogram Bar Bar w Error Dot Plot 2D Time Serie QQ Plot Contour Lower X Axis Left Y Axis No Conditioning Plots in Separate Panels Line High Density Color Plot Smoothing Spline Polynomial Fit BoxPlot Density Grouped Bar Grouped Bar w Error H orizontal Bar H igh low Average Area Filled Contour Upper X Axis Right Y Axis 4 Panel Conditioning Scatter w Line Line w Text as Symbols Bubble Color Robust Fit Exponential Fit Pie H istogram D ensity Stacked Bar Bar Origin Base Stacked H orizontal Bar Vertical Error Bar Scatter Plot M atrix Levels Plot Upper X with Frame Right Y with Frame 9 Panel Conditioning Sep Panels w Varying Y axis Sep Panels w Varying X axis Figure 2 12 The 2D Plot Palette see graphics section starting with chapter 8 Creating a Graph 52 S PLUS PALETTES AND TOOLBARS Plots3D Palette Scatter D rop Line Scatter Coarse Surface Coarse Filled Surface 8 Color Draped Surface B
163. Rotation 4 Pane Rotation 6 Pane Rotation Condition on X Condition on Y Condition on Z No Conditioning 4 Paned Conditioning 6 Paneg Conditioning Figure8 3 The3D Plot Palete Creating a Graph You can also create a graph by dragging and dropping each component of a Using Drag and graph onto a Graph sheet Drop To create a graph using drag and drop 1 Open aD ata window or O bject Browser containing the data to plot 2 Create anew Graph sheet or open an existing Graph sheet 3 From the Window menu choose Tile Vertical Now you can see the data and Graph sheet simultaneously Select the Graph sheet window by clicking in its title bar 4 Click the 2D or 3D Plots button on the toolbar A palette of 179 CHAPTER 8 CREATING A GRAPH available plot buttons appears 5 Drag the desired plot button from the palette and drop it inside the Graph sheet D efault axes are drawn and a plot icon is drawn on the graph 6 Select the data columns you want to plot Use cTrt click to select discontiguous columns 7 Position the mouse within the selected region not in the column header until the cursor changes to an arrow Pressing the left mouse button drag the data and move it over a plot icon When the plot icon changes color release the mouse button to drop the data and generate the plot 8 Theplot icon is replaced by an actual plot of the data columns Figure8 4 An S PLUS plot in this casea Linear Fit Creating a
164. SPSS files SAV SAS files SD 2 SAS Transport files version 6 x TPT Some special export options may need to be specified in your SAS program We suggest using the SAS Xport engine not PROC CPORT to read and write these files M icrosoft Access files M D B M atlab M AT SPSS Export files PO R S PLUS Transport Files SD D STATA files D TA Versions 2 0 and higher Gauss DAT files automatically reads the related DHT file To import a data file 1 From the File menu choose Import D ata then From File 2 In the File name box type or select the name and optionally the path of the file you want to import 3 In the Files of type box select the type of the file to import 4 In the Import To box specify the target data frame starting column and whether you want to preview the imported file first THE DATA IMPORT DIALOGS 5 Click O pen to import the file Note If a file extension is inappropriate an error may appear indicating an unrecognized format or the data file may be converted incorrectly The Data Import Dialogs When you are importing most file types typically you only need to specify the file name and file type and the file will be imported into a new data frame and opened in a D ata window using default settings You can specify your own settings on the O ptions and Filter pages of the Import D ata dialog All three pages share a common lower section T he
165. Show Menu In Pus in the context menu for the Menultem object which represents the default S PLUS menus Menu items are easily copied moved and deleted through the O bject Browser Moving Menu Items To move a menu item into a different menu locate the menu item icon in the O bject Browser Select the icon hold down the ALT key and drag it onto the menu to which it is to be added To move the menu item within its current menu hold down the SHIFT key and drag the menu item icon to the desired location MENUS Saving and Opening Menus Example Customizing the Context Menu Copying Menu Items To copy a menu item into a different menu hold down the CTRL key and drag its icon onto the menu to which it is to be added To copy a menu item within its current menu hold down the SHIFT and CTRL keys and drag the menu item icon to the desired location Deleting Menu Items To delete a menu item right click on the menu item in the O bject Browser and select D elete from the context menu To save a menu to an external file right click on the M enultem object in the O bject Browser and select Save As in the context menu Enter a filename in the Save As dialog and click OK The extension SMN is added to the filename To open a menu which has been saved in an external file right click on the default M enultem object and select Open from the context menu In the Open dialog navigate to the desired file select it and c
166. T he selected plot is redrawn using the new type If your data are not appropriate for the chosen plot type the plot appears on the graph in an iconized form To close the palette click the close box or click the plots button on the standard toolbar again To change the plot type using a dialog 1 2 3 Select the plot and choose C hange Plot Type from the Format menu In the Change Plot Type dialog choose from the list of plot types available for the current axes type for example 2D As you click through the list of plot types sample plots are displayed in the dialog Choose OK The plot is redrawn using the new plot type Table 12 1 outlines the allowed plot types CHANGING THE PLOT TYPE Table 12 1 Legal plot types Plot Type 2D Axes 3D Axes Matrix Axes Polar Axes Text Axes Area Bar Bar H oriz Bar 3D Box Bubble Bubble Color Color Comment Contour Contour 3D Density Dot Error Bar Fit Exponential Fit Linear Fit Log 10 Fit Loge Fit Polynomial Fit Power High Density Error Bar H oriz ee 1 297 CHAPTER 12 EDITING PLOT PROPERTIES Table 12 1 Legal plot types High Low Plot Histogram Levels Plot Line Line 3D Pie QQ Plot Regression 3D Robust Line Scatter Scatter 3D Scatter Plot M atrix Smoothing Friedman Super Smoothing Kernel Smoothing Loess Smoothing Spline Step Plot Surface Time Ser
167. THE LICENSE FEE ACTUALLY PAID BY LICENSEE TO MATHSOFT E SOME STATES DO NOT ALLOW THE EXCLUSION OF IMPLIED WARRANTIES SO THIS LANGUAGE MAY NOT APPLY IN SUCH CASE MATHSOFT LIABILITIES WILL BE LIMITED BY THE ABOVE LIMITATION OF REMEDIES PROVISION F Lucent Technologies does not warrant the Software does not assume any liability 619 LICENSE AGREEMENT AND LIMITED WARRANTY VI Indemnity by Licensee to MathSoft Vil Termination Vill Miscellaneous 620 regarding the Software and does not undertake to furnish any support or information regarding the Software A Licensee indemnifies and holds harmless MathSoft from any and all claims demands or actions based on or relating to Clients or to services offered by Licensee involving use of the Software or Clients or based on representations or statements made by Licensee or its agents or other actions of Licensee or its agents A The license granted under Articles II shall remain in force unless Licensee breaches any material term of this Agreement in which case MathSoft shall have the right to terminate these licenses Regardless of whether these licenses expire or are terminated all other articles of this Agreement shall survive perpetually B Upon the termination or expiration of the licenses granted under Articles II all rights granted to Licensee will terminate and revert to MathSoft and Licensee promptly must delete and destroy all copies of the Software and retur
168. THE SCRIPT WINDOW To find text g 1 Click the Find button on the Script toolbar or choose Find from the Edit menu or click on CTRL F The Find dialog will pop up 2 In the Find What box type the text you re searching for Find Find what M Match whole word only Cancel Match case Figure 21 5 TheFind dialog will find a string of up to 255 characters the text will scroll hori zontally as you type If you used Find or Replace in your current work session the text you last searched for is selected in the Find W hat box Type over the text to find different text 3 Choose the Find N ext button to begin searching The Find dialog has the options in Table 21 2 Table 21 2 Check box optionsin the Find and Replace dialogs Option Purpose M atch whole word only Choose this option to find whole words not substrings M atch case Choose this option to find only words having the specified pattern of uppercase and lowercase letters To find and replace text 1 From the Edit menu choose Replace or type CTRL H The Replace dialog will pop up 2 In the Find W hat box type the text you re searching for 553 CHAPTER 21 WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS 3 If you used Find or Replace in your current work session the text you last searched for is selected in the Find W hat box Type over the text to find different text Replace Find what mean Eind Next Rep
169. These remedies shall not be deemed to have failed of their essential purpose as along as MathSoft is willing to take one of those actions A The warranties are being provided only to the original Licensee no warranties of any kind are provided to any other parties B The warranties do not cover damage or defects caused by or related to misuse accident negligence or misapplication Because programs such as this are inherently complex MathSoft does not warrant that the Software is error free or will operate without termination Furthermore MathSoft does not warrant that the Software will work with any given database network or network application C MathSoft hereby warns Licensee that due to the complexity of the Software it is possible that use of the Software unintentionally could lead to the loss or corruption of data Licensee assumes all risk for such data loss or corruption the warranties provided hereunder do not cover any damage or losses resulting therefrom D IN NO CASE SHALL MATHSOFT BE LIABLE FOR ANY INCIDENTAL SPECIAL OR CONSEQUENTIAL DAMAGES OR LOSS INCLUDING WITHOUT LIMITATION LOST PROFITS OR THE INABILITY TO USE EQUIPMENT OR ACCESS DATA WHETHER SUCH DAMAGES ARE BASED UPON A BREACH OF EXPRESS OR IMPLIED WARRANTIES BREACH OF CONTRACT NEGLIGENCE STRICT TORT ORANY OTHER LEGAL THEORY THIS IS TRUE EVEN IF MATHSOFT S ADVISED OF THE POSSIBILITY OF SUCH DAMAGES IN NO CASE WILL MATHSOFT S LIABILITY EXCEED THE AMOUNT OF
170. To zoom a Graph sheet 1 From the View menu choose Zoom 2 From the Zoom menu choose the desired zoom percentage You can use all normal editing and formatting options on the zoomed Graph sheet 3 When you are finished choose Fit in Window from the Zoom menu to return the Graph sheet to its original size 4 Graph sheets are always printed at 100 size even if they are zoomed Viewing in Draft Use Draft mode to increase the display speed of Graph sheets You can toggle Mode this option on and off To view a Graph sheet in Draft mode w From the View menu choose Draft or click on the Set Draft M ode toolbar button Draft mode only affects screen resolution all output will be publication quality Note You can change the default settings for D raft mode in the O ptions Graph O ptions dialog Viewing Full You can view Graph sheets at full screen size without the menu bar window Screen title bar or toolbars To view a Graph sheet Full Screen 1 Press F2 or choose Full Screen from the View menu 2 Click the mouse or any keyboard button to return to the original view 182 VIEWING GRAPH SHEETS Adding a Plot to a Graph 3 oO E o 2 E Plots can be added to an existing graph using the plot buttons or the menus o Treatment a m Treatment B 20 Treatment B a fo D
171. Up M oves to first row current D rag vertical slider to top arrow D rag vertical slider to bottom arrow Click the column header SELECTING CELLS COLUMNS AND ROWS Table 4 2 Keyboard and mous shortcuts SH IFT Spacebar CTRL SH IFT Spacebar SH IFT Arrow Keys Fl F5 F9 Using the Go To Cell Option Selects a row Click the row number Selects the entire D ata window Click the upper left hand corner of the D ata window Puts cursor in selection mode D rag the mouse across cells M ove cursor to make block selection Displays on line help Click the H elp button DisplaysGo To Cell dialog when Choose View Go To Cell a D ata window is selected Edits the column name when a D ouble click in the column name D ata window is selected portion of the header W hen entering data you can use the mouse scroll bars direction keys or the Go To Cell option to move from cell to cell in a D ata window T he cell that the cursor occupies is always highlighted with a box around it Use the BACKSPACE and DELETE Keys to erase typing errors and the mouse or RIGHT and LEFT direction keys to move around in the entry To cancel an entry when typing press Esc You can press ENTER UP or DOWN to enter the data value in the highlighted cell If you press ENTER the value is entered in the cell and the cursor moves to the next cell S PLUS s smart cursor feature moves the cursor in the direction of the las
172. a Formatting Panels Formatting 2D Axes Formatting 2D Axis Labels Formatting and Rotating 3D Axes Formatting 3D Axes Labels Formatting 3D Planes Rotating a3D Graph Displaying 3D Multipanel Graphs Formatting Polar Axes Adding Multi line Text Adding Special C haracters and Formatting Text Adding Titles and Legends Adding 2D Axis Titles Adding 3D Axes Titles Adding aD ate and Time Stamp Adding a Legend Formatting the Legend Items Adding Labels for Points Adding a Curve Fit Equation Adding Lines Shapes and Symbols Summary of the Annotation Palette 228 233 233 234 238 239 244 249 250 251 254 256 259 261 265 267 268 268 269 270 271 273 274 275 277 227 CHAPTER 10 FORMATTING A GRAPH FORMATTING A GRAPH SHEET Layout Page 228 In S PLUS you have complete control over the format of your graph including axes plots titles and comments See chapter Creating a Graph for step by step instructions for creating graphs Once you have created a graph you can take advantage of S PLUS s many editing options The rectangular page in the window is the Graph sheet To format the Graph sheet 1 With the Graph sheet in focus choose Sheet from the Format menu The Graph Sheet Properties dialog appears or right click outside of the page to bring up the shortcut menu 2 Make the desired formatting changes to the Graph sheet 3 ChooseOK In the Graph Sheet Properties dialog the L
173. a s Guide to further acquaint themselves with the graphical user interface N ew features and techniques include Statistics Bootstrap and jackknife estimation Linear mixed effect models Nonlinear mixed effects M ultiple comparisons Crisp and fuzzy clustering Monothetic clustering e Divisive and agglomerative methods Graphics Multiple simultaneous 3D rotation views e 2D projectionsin 3D space Multiple graphs per page with auto formatting Interactive 3D view angle specification Drag and drop creation of Trellis graphics Point and click editable graphics Flexible page layout 23 CHAPTER 1 WELCOME TO SPLUS M ultiple axis breaks M ultiple line text annotation Superscripts subscripts Greek letters and symbols Automatic creation of code to produce editable graphics Support for international ASCII character set User Interface 32 bit application with full Windows 95 compatibility also compatible with Windows NT and 3 1x running W in 32s M icrosoft O ffice compatible user interface Statistical menus and dialogs Editable graphics Customizable toolbars menus and dialogs Tooltips online help and tutorial D ata windows for spreadsheet editing and display of data M ulti page Graph sheets O bject Browser to organize data graphs functions and other objects Script editor for scripting and programming Report windows for easy handling of output Import and Export 24 Import SAS
174. a skip You may need to skip characters when you want to avoid importing some characters in the file For example you may want to skip blank characters or even certain parts of the data If Preview Only is specified for a formatted ASCII file but no format is given S PLUS still tries to display the data in a readable way T he field width is set equal to the default column width and all columns are left justified On the screen the file appears as it would in an ASCII editor If there is a line feed at the end of the start row up to a maximum length of 4096 characters the length of the row is set equal to the actual value thus determining the number of columns to read in If you wish to import only some of the rows specify a starting and ending row If each row ends with a new line S PLUS will treat it as a single character wide variable that is to be skipped 129 CHAPTER 5 IMPORTING AND EXPORTING DATA IMPORTING ODBC TABLES Introduction to ODBC Installing ODBC 130 The ODBC dialog controls the import of tables from databases and applications that support the O pen D ataBase C onnectivity specification You may use a Filter specification to subset the data to be imported By default an entire table is imported to an S PLUS data frame Applications such as M icrosoft Access and Excel as well as most commercial databases generically known as data sources support the Open DataBase Connectivity O D BC standard
175. a Has a Grouping Variable Check this if one column in the data frame is a grouping indicator that categorizes cases into two groups In this case select the response from Response Variable and the indicator from Grouping Variable Data Frame Select a data frame x Variable Select a column as the firs sample when the data frame does not have a grouping indicator y Variable Select another column as the second sample when the data frame does not have a grouping indicator gt Tip You can have a data frame containing a respons variable and a grouping variable or a data frame with two samplesin two columns 392 TWO SAMPLE KOLMOGOROV SMIRNOV TEST Results Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last Im is the most recent linear regression model fit Print Results Check this to print the results of the Kolmogorov Smirnov test in the designated output window Related S PLus language functions ks gof chisq gof qqplot 393 CHAPTE
176. a dialog 119 D ata Specs page 120 Filter page 123 Options page 121 importing data 117 importing graphics 164 importing dBase files 128 importing Lotus files 127 Increase Precision button 115 increment multiple 334 in place activation 162 Insert Picture option 165 inserting blocks 103 graphs 190 rows 103 inserting columns and rows 102 installing on anetwork 17 installing the software 16 irregular form data 200 INDEX J justifying comment plot text 310 pie chart labels 326 K kernel bandwidth 375 kernel smoother 374 kernel specifications 374 keyboard and mouse shortcuts 92 knots 378 K olmogorov Smirnov test 384 two sample 392 K ruskal W allis test 396 L label display format pie charts 325 label frequency contours 312 labels contour plots contour plot labels See labels pie chart slices pie chart labels See levels column 313 levels contour 313 line color 301 line plots 319 lines 319 presorting data 321 line plots 3D colors 331 data specifications 330 plot options 330 presorting data 330 symbols 330 linestyle 301 line styles 603 line weight 301 linear regression 414 data specifications 313 formula building 536 Linear Regression dialog 78 link function 458 linking data 166 Links dialog 167 lists 88 locally fitted polynomial 371 locally weighted regression 370 433 logistic regression 426 log linear regression 427 long form stacked data 199 Lotus files 127 low frequency emphas
177. a frame gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame Weights Enter the column that specifies weights to be applied to all observations used in fitting the tree To weigh all rows equally leave this blank 473 CHAPTER 18 FITTING STATISTICAL MODELS 474 Formula Fitting Options Subset Rows with Enter an S PLUS expression which identifies the rows to use in fitting the tree To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values T RUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the modal If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Formula Enter a formula specifying the desired model A valid formula has the following form response predictorl predictor2 Create Formula Click this to open a formula
178. a mirror of the opposite axis and always has identical scaling Select the graph From the Insert menu choose Axis From the Axis submenu choose X Axis with Frame or Y Axis with Frame T he Axis dialog appears 239 CHAPTER 10 FORMATTING A GRAPH 4 Specify the position and formatting attributes for the new axis and choose O K The axis and frame are added to the selected graph Formatting a 2D To select an axis Axis Display Scale Page 240 w Click inside the tick area of the axis W hen the axis is selected it will have a square green knob in the center of the axis If you see a triangular green knob you have selected the axis labels not the axis To move an axis and offset it from the plotting area w Select the axis A knob appears in the center of the axis Drag the knob until the axis isin the desired position or w Doubleclick the axis or select the axis and choose Selected Axis from the Format menu Choose the D isplay Scale page and specify a value in the Axis O ffset field Choose OK To format an axis 1 Doubleclick the axis or select the axis and choose Selected O bject from the Format menu 2 From the Axis dialog choose the desired page D isplay Scale R ange Grids Ticks or Axis Breaks M ake the changes and chooseO K Alternatively you can access pages of the property dialog by selecting the axis right clicking and selecting a page from the shortcut menu In the Axis dialog the D isplay
179. aced in the column list field Press TaB to move to another item in the dialog or press ENTER to close the dialog and execute the command Note Only column names are displayed in the field menu lists not column numbers If your columns are not named you will need to specify the column number instead 95 CHAPTER 4 USING DATA WINDOWS ENTERING DATA Entering Data from the Keyboard 96 When you open a new Data window it is formatted using default settings The cell that the cursor isin is highlighted N umeric data is the default data type for the initial columns To enter other types of data you need to insert columns of the appropriate type or convert numeric columns to character To set a right margin When entering or editing data you can set the right margin for cursor wrapping in the D ata window 1 Position your mouse in the desired column and right click to display the Column shortcut menu 2 From the shortcut menu choose Set Right M argin If you arein column 5 when you set the margin column 5 becomes the new right margin As soon as you enter data in column 5 and press ENTER the cursor wraps to column 1 in the following row You can have a different margin setting for different views on the same data set M argin settings are retained until you close the Data window or exit S PLUS To enter data 1 Click in the cell in which you want to enter data 2 Typein the data value using the
180. acters and can be any combination of letters numbers symbols or spaces The column description is used as the default axis title in your graph For example if your Y column has a description for example Rates of Change it is used for the y axis title You may leave the description blank If there are no column names or descriptions S PLUS uses the column numbers for the default legend text To change a column name or description 1 Select the column and choose Selected O bject from the Format menu 113 CHAPTER 4 USING DATA WINDOWS To Change the Data Type of Columns To Change Display Format and Precision of a Numeric Column 114 2 Type the desired name in the Column Name field and optionally type a Description in the Column Description field 3 ChooseOK To edit a column name 1 Doubleclick in the column name portion of the column header or press F9 2 Type the desired name or make any desired modifications to an existing name 3 Press ENTER or click elsewhere in the Data window to accept the changes In S PLUS there are several data types You can easily convert between the data types using the Change D ata Type button or the change option on the D ata menu To change a column type 1 Select the column 2 Click the Change D ata Type button see Table 4 1 T he D ataFrame toolbar or choose C hange D ata Type from the D ata menu T here are several types of formats for numeric column data
181. ady exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame This allows you to keep predicted values from a model with the original data or to keep the predictions from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Predictions Check this to save predictions to the data frame specified in Save In Related S PLus language functions for Principal Components Analysis princomp princomp object loadings bi plot princomp screeplot plot loadings 504 PRINCIPAL COMPONENTS ANALYSIS Other related S PLus language functions svd cancor factanal 505 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS NONPARAMETRIC SURVIVAL This dialog computes the estimate of a survival curve for censored data using either the K aplan M eier or the Fleming H arrington method See also chapter 22 Overview of Survival Analysis and chapter 23 Estimating Survival in the Guide to Statistics To perform nonparametric survival modeling Choose Statistics gt Survival gt Kaplan Meier from the main menu The dialog shown below appears Hepner Gerri Fill E Moda Options Pesis me Oma Model Deis Fano awT iapiarmit 7 Wiegi
182. aiis Preamen Coevengence Tolerance rin Pret et aizen a Maimun harasa e Options Page Use the Options page to define optimization parameters for model Optimization Parameters computations Convergence Tolerance Enter anumber specifying the convergence tolerance Iteration will continue until the relative change in deviance is less than this number Initial Parameter Values Enter a vector of initial values If this is left blank zero will be used for each variable Maximum Iteration Enter a number specifying the maximum number of iterations If convergence has not been reached after this number of iterations the procedure will stop 523 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS Paranentic mail Fila Ei Modal Optore Aenubs Priel Fiale Saud Raris TD Ehi Guaput Save in F Long Outpt r Fitted aloe M Driene Roniduse r Fanan Reaatuels P Wetorang Aenda T kiai Plesichu als Results Page Printed Results Short Output Check this to print a summary of the model results in the designated output window T hisincludes estimates of the coefficients dispersion scale degrees of freedom and 2 loglikelihood Long Output Check this to print a long summary of the model results in the designated output window This includes summary statistics for the deviance residuals standard errors and z values for the coefficients number of iterations and correlation of coefficients Saved Resul
183. ain menu The dialog shown below appears Aaadom Humba Goneration Big Ei Sonple Sire Pamen of the Deeded Destibuticn kini fo SetSeed ah Mason urn fl Cis bution urniks ih Parameters ofthe Selected Diritti Piemslie Ean ii issida E Priel Fewelts Sample Size Enter the sample size Set Seed with Enter a random number seed used in the random generation algorithm When this field has a value rerunning the dialog in the same state will reproduce the same data Distribution Select the distribution from which to generate the random sample RANDOM NUMBER GENERATION Parameters of the Selected Distribution Results Enter the parameters of the hypothesized distribution For example if the Cauchy distribution is chosen from the Distribution drop down box Location and Scale are highlighted Enter the location and scale parameters of the C auchy distribution Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last The default is last rdist Print Results Check here to prin
184. al values each given in the form name value Every parameter to be estimated must appear both here and in Formula For example Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names T he default is last nls T he saved object will be of class nls See the on line help for nls object for more information about the saved object Howlinoer Lowst Squatos Regression Model Optom Penia Predial scar al aih zi P Lia Partial Lit as Algom Rimore armiga foU Comenercs Tosrerce 10r hin Scabs tor Sip Sarintage po 1 l Print Ension Trace Use the Options page shown above to control the way in which the nonlinear least squares regression is carried out Maximum Iteration Enter the maximum number of iterations to allow during fitting Convergence Tolerance Enter a positive number used as the tolerance for the convergence criterion in the algorithm This relative offset criterion measures the numerical imprecision in the parameter estimates compared to the statistical variability NONLINEAR LEAST SQUARES REGRESSION Results page Printed Results Sav
185. alog for the function can be made to appear OPEN The file specified on the Command page is opened when the button is clicked The file will be opened by the application associated to it by the operating system PRINT The file specified on the Command page is printed when the button is clicked The file will be printed by the application currently associated to it by the operating system RUN Thefile specified on the Command page is opened and run as a script by S PLUS when the button is clicked Tip Text Enter tool tip text for the button Hide Check this to make the button invisible W hen the item is hidden its icon in the O bject Browser appears grayed out This can also be specified through the context menu D eletable Check this to allow the item to be deleted M ove to the Command page of the dialog Built In Operation Select from the list of actions associated to the built in S PLUS menus and toolbars It is not possible to easily modify this list Command Enter the name of an S PLUS function or path and filename This field is enabled when Action is set to FUNCTION OPEN PRINT or RUN on the button page Use the Browse button to identify the folder Parameters Edit the text in this field to specify the arguments for the function which will execute when the button is clicked The easiest way to specify these arguments is to work through the Customize dialog available through the context menu for the item in the
186. alog shown below appears Paranentic Barsil Bll Moda ptora Fiesub Daia Mode Oats Frame Chirmizarion aiene Subed Aras wih Link og F F Ont Foes e ng Values Food Poesmnetere Types of Gereanng jag ha Saw Modd Obpact Save ie fisaicur J Foemuls Fariba Crosio Fomula Model Page Data Data Frame Select a data frame gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame Weights Enter the column that specifies weights to be applied to all observations used in the regression To weight all rows equally leave this blank 521 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS Formula Model Save Model Object 522 Subset Rows with Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any
187. alues Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the model If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Formula Enter a formula specifying the desired model Create Formula Click this to open a formula builder dialog used to construct a formula specifying the desired model See the chapter Building Formulas for more information Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last The default is last aov n u The saved object will have class aov maov or aovlist See the on 445 CHAPTER 18 FITTING STATISTICAL MODELS line help for aov object for more information about the saved object Fined CHocts Anabasa of wananga mE Model Optra Penta Pine Conas Auaign Conant ip vansa Comins Options page Contrasts Assign Contrast Choose contrasts for the
188. ames Choose whether to include column names in the file Quote Char Data Choose whether to have quotation marks used to enclose character data in the file e g List 1 Line Length The standard line length is 80 characters per line based on 8 5 by 11 inch paper A new line is started after 80 characters have been printed If you increase the line length characters may run off the page when printed or viewed in a text editor The line length specification is only used when nothing is specified for the D elimiter 137 CHAPTER 5 IMPORTING AND EXPORTING DATA EXPORTING ODBC FILES The ODBC dialog controls the export of data sets to ODBC databases and applications that support the O pen D ataBase Connectivity specification By default an entire data set is exported to an ODBC table Exporting Data 1 From the File menu choose Export Data then To ODBC Sets to an Connection External ODBC 2 From the ODBC Data Source dropdown list select the data source Data Source to which you wish to export a data set If the desired data source is not available or the list is blank you may create a data source by selecting the button labeled to the right of the ODBC Data Source list box 3 Select the name of the data set you wish to export in the Export From group A default table name will appear in the Table Name field 4 Click OK to export the data set and create a table ligai Kanani Teaia Ehet q a cc J cee
189. amp Brooks C ole Pacific Grove CA Spector P 1994 An Introduction to S and S PLus Duxbury Press Belmont CA Data Analysis Bruce A and Gao H Y 1996 Applied Wavde Analyss with S PLUS Springer Verlag N ew York Chambers J M and Hastie T J 1992 Statistical M oddsin S Wadsworth amp Brooks Cole Pacific Grove CA Everitt B 1994 A Handbook of Statistical Analyses Usng S PLUS Chapman amp Hall London 21 CHAPTER 1 WELCOME TO SPLUS 22 H ardle W 1991 Smoothing Techniques with Implenentation in S Springer Verlag N ew York Kaluzny S P Vega S C Cardoso T P and Shelly A A 1997 S SPATIALSTATS User SM anual Springer Verlag N ew York M arazzi A 1992 Algorithms Routines and S Functions for Robust Statistics Wadsworth amp Brooks Cole Pacific Grove CA Venables W N and Ripley B D 1994 Modern Applied Statistics with S PLUS Springer Verlag N ew York Graphical Techniques Chambers J M Cleveland W S Kleiner B and Tukey P A 1983 Graphical Techniques for D ata Analyss Duxbury Press Belmont CA Cleveland W S 1993 Visualizing D ata H obart Press Summit NJ Cleveland W S 1985 The Elements of Graphing Data Hobart Press Summit NJ WHAT S NEW IN S PLUS WHAT S NEW IN S PLus New Features T he following is a summary of new features in S PLUS Users of S PLUS 3 3 for Windows can browse the rest of this Us
190. and Ozon 1850 HJY 0 200 100 E m 3 ae a S TD cc FE i FELEK Zago 0 a0 100 100 z fo 130 189 Linon Lona nial on Figure 3 23 Ozone and Solar Radiation notice that the rdationship is stronget when Temperature is high and Wind Speed is low 4 Select the plot by clicking on the line or symbols in any one of the panels O pen the 2D Plot palette and click on the Linear Fit button A linear regression line will replace the loess curve in each panel 5 Save the multi panel plot as mpanel sgr using File Save As and 76 IN DEPTH TOUR Solar Raat on Close the D ata window Graph sheet and 2D Plot palette The Relationship Between Radiation and ozone ja S70 tn hao TERPERATURE sid HD 00 _ RADLTION 100 2 Oto rai TEAST LR 2 fo 130 129 Chana oonmantraton Figure 3 24 A linear regression has replaced the loess curvein each pand Applying Statistics Models S PLUS provides a vast array of statistical techniques with the most widely used techniques accessible through dialogs launched from the Data and Statistics menus All techniques are available through the S language Commands may be issued interactively in the Commands window or as a script in a Script window In the course of an analysis the user may begin by fitting a model through a convenient dialog then proceed to analyze the mode and perform diagnostics through the flexible and powerful S language In this se
191. and Response vs Fit plots See the on line help for oess smooth for details Include Rugplot Check this to display a rugplot on the Residuals vs Fit Sqrt Abs Residuals vs Fit and Response vs Fit plots A rugplot is a sequence of vertical bars along the x axis that mark the observed x values Number of Extreme Points to Identify Enter the number of extreme points that will be identified on the Residuals vs Fit Sqrt Abs Residuals vs Fit Residuals Normal QQ and Cook s Distance plots The row names from the data frame specified on the model page will be used to identify the points Plot Partial Residuals Check this to display partial residual plots for all the terms in the modal Include Partial Fit Check this to display the partial fit for the term Include Rugplot Check this to display rugplots on the partial residual plots A rugplot is a sequence of vertical bars along the x axis that mark the observed x values 7 Common Y axis Scale Check this to give all the partial residual plots the same vertical units T his is GENERALIZED LINEAR MODELS essential for comparing the importance of fitted termsin additive models Gesemirod Lingpa Models Mock Opkoara Piera ue Fig Pes Liste oe E Pradicion Type mi Seren in l Fricis P Sardan Errore Predict Page New Data Enter the name of a data frame to use for computing predictions It must contain the same names as the terms in the right side of the f
192. and the columns to sort by To sort data using the toolbar 1 Select the data columns you wish to sort by All columns are sorted 2 Click the Ascending Sort button or the D escending Sort button The data columns are sorted in ascending order or descending order starting with the last column that was selected To sort data using the Sort Columns dialog From the D ata menu choose Sort 1 In the Sort dialog you have the following options Sort Column s Specify the names or numbers of the columns to be sorted and the appropriate data set To sort all of the columns in a data set enter ALL Sort By Column s Enter the name or column number of the columns to sort by for example GENDER AGE If you list more than onecolumn the data in the specified SORTING DATA columns will first be ranked according to the first column specified Then in the case of equivalent data the second column will determine the ranking and then the third column and so on The columns listed here do not have to be the same as the columns being sorted For example you can sort INCOM E and AGE by GENDER Also these columns need not be in the same data set Ascending Check this box for ascending order otherwise the sort is in descending order Target You have the option of specifying a target Data window and column for storing your sorted data If you do not specify anything in these fidds sorting will be done in place and your
193. aneously Select the Graph sheet window by clicking in its title bar 4 Click the 2D or 3D Plots button on the toolbar A palette of available plot buttons appears 5 Drag the desired plot button from the palette and drop it inside the Graph sheet D efault axes are drawn and a plot icon is drawn on the graph 6 Select the data columns you want to plot Use cTRL click to select discontiguous columns 7 Position the mouse within the selected columns until the cursor changes into an arrow Pressing the left mouse button drag the data and move it over a plot icon When the plot icon changes color release the mouse button to drop the data and generate the plot You can easily combine graphs from different Graph sheets onto one Graph sheet S PLUS arranges them on the page automatically using a default Arrange Graphs setting See the following section for more information on arranging graphs after they are combined on one Graph sheet 191 CHAPTER 8 CREATING A GRAPH Arranging Graphs on the Graph Sheet 192 To combine graphs from different Graph sheets on to one Graph sheet Load the Graph sheets containing the graphs you wish to combine 1 Click the New button and choose GraphSheet to create an empty Graph sheet 2 Choose Tile Vertical from the Window menu 3 Select one of the Graph sheet windows by clicking in its title bar 4 From the Edit menu choose Copy GraphSheet This copies all objects in the current Gra
194. anges color release the mouse button to drop the data and generate the plot The plot is redrawn using the new data If only one column is dropped on a plot requiring two columns of data it replaces the Y column of data If multiple columns are dropped multiple plots of the same type are created For all plots you must specify one or more columns of data to be used when plotting If you select data columns before choosing a plot from the palette or before choosing Graph from the Insert menu the column names and data set name are filled in for you in the plot property dialog O therwise you can use this dialog to specify the data set and desired columns Alternatively you can enter an equal sign with an S PLUS expression resulting in a vector See section Selecting Cells Columns and Rows page 90 for information on selecting columns of data To edit data specifications for a plot 1 Doubleclick on the plot or plot icon or select the plot and choose Selected Plot from the Format menu or right click on the plot and Select Data to Plot from the menu that appears A plot property dialog will appear 2 On the Data to Plot page specify the columns necessary for that particular plot type See Table 8 2 in the section Preparing D ata for Graphing page 194 to determine which columns need to be specified for a particular plot type Data Set The data frame list matrix or vectors containing the data columns to be plotted Click
195. anging D efaults and Settings 598 Saving O bject D efaults 598 Specifying G eneral Settings 598 Specifying Command W indow Settings 600 Specifying Undo amp History O ptions 600 Specifying Text O utput Window Settings 601 Specifying Graph O ptions 601 Specifying Graph Styles 602 Specifying C olor Schemes 603 Redrawing Plots Automatically 604 In S PLUS you have many options for customizing your workspace You can specify the defaults for any object choose whether to display toolbars and set graph redraw options In addition you can create a default project with D ata windows Graph sheets and scripts You can further customize entire work sessions using customizable menus and toolbars see the previous chapter Customizing the U ser Interface 597 CHAPTER 23 CUSTOMIZNG YOUR S PLUS SESSION CHANGING DEFAULTS AND SETTINGS Saving Object Defaults Specifying General Settings General Page 598 In S PLUS you can specify the defaults for any object including symbols plots titles graph sheets and data objects To save the defaults for an object 1 Create the object using the exact specifications you want to save as the default For example create a symbol and specify the fill color and size you want to use as the default 2 Select the object if it is not already selected 3 From the Options menu choose Save O bject as D efault The name of the selected object will replace the word O bject in the menu optio
196. ar XY PlaneZ Min XY PlaneZ M ax 2 Panel Rotation Condition on X No Conditioning Line Regression D ata Grid Surface D ata Grid Filled Surface 16 Color Draped Surface Contour XZ PlaneY Min XZ PlaneY M ax 4 Panel Rotation Condition on Y 4 Panel Conditioning Line with Scatter Regression with Symbols Spline Surface Filled Spline Surface 32 Color Draped Surface Filled Contour YZ PlaneX Min YZ PlaneX Max 6 Panel Rotation Condition on Z 6 Panel Conditioning Figure 2 13 The3D Plot Palette see section on graphics starting with chapter 8 Creating a Graph 53 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE Annotation Palette Annotation Sdection Tool Line Tool Radial Lines Tool Filled Box Shape Square X X Box Filled circle Box down triangle Dash Comment Tool D ata Stamp Tool Error Bars Tool Box Shape Circle Diamond Diamond X Circle Filled up triangle X Diamond Bar Label Point Tool Arrow Tool Vert Ref Line Rounded Box Up triangle D own triangle Circle Box up triangle Filled diamond Cross M ale Select Row Tool Arcs Tool H oriz Ref Line Ellipse Shape Plus Box X Up Down triangles Filled square Filled down triangle Ant Female Figure2 14 The Annotation toolbar has buttons for Tools Shapes and Symbols See chapter 10 Formatting a Graph 54 TUTORIAL Quick Tour 57 In D epth Tour 65 The O bject Browser 65 I
197. ar regression example 539 ovarian data frame example 542 Remove group box 537 Special T erm group box 537 Survival Response group box 542 transformation 538 transformation Function group box 538 transformation Function with Constant group box 539 transformation Variable group box 538 Variable group box 536 609 INDEX frequency symbol 302 Friedman test 398 fuel frame data frame 57 225 G galaxy data frame 61 222 General Settings dialog 598 generalized additive model 465 generalized linear model 456 GoTo Cdl dialog 94 graph objects dragging into scripts 560 zooming 182 Graph O ptions dialog 601 graph sheets opening existing 175 overview 174 Graph toolbar 175 graphics importing 164 graphs exchanging 193 multipanel 210 overview 174 preparing data 194 gridded contour plots 311 grids 334 grouped bar charts 304 603 offset 306 grouped box plot 308 grouping indicator 388 390 392 H handling missing data 366 height symbol 302 H elp system On line Demos 19 on line help 19 On Line M anuals 19 training courses 20 610 hiding plots 186 highlighting brush and spin 204 high low average plots 317 data specifications 317 plot options 317 histograms 318 History log 558 entries 600 undoing actions 109 history log clearing 559 executing commands 559 size 558 History options 600 history rollback 109 horizontal error bars 316 hypothesized distribution 356 359 385 image colors 604 Import Dat
198. as Symbols Q e Q 195 CHAPTER 8 CREATING A GRAPH Table 8 2 Required columns for different plot types required O optional Line 3D Q Q e Line with Scatter 3D Q QO e Pie e Polar radius angle Projections 3D Q Q OQO Quantile Q uantile Plot y1 y2 e QQ Normal with Line y e Regression 3D QO e o Regression and Scatter 3D O e e Robust Least Trimmed Squares O e Robust M M Q e Scatter Plot QO e Scatter 3D Q QO e Smoothing Friedman Super Q e Smoothing Kernel QO e Smoothing Loess O e Smoothing Spline Plot Q e Step Plot H orizontal QO e Step Plot Vertical Q Surface C oarse Grid Q Q Surface D ata Grid Q Q e Surface Spline Fine Grid Q Q o Surface Filled C oarse Grid Q Q e 196 PREPARING DATA FOR GRAPHING Table 8 2 Required columns for different plot types required O optional Surface Filled D ata Grid Q Q e Surface Filled Spline Fine Q Q e Grid Surface 8 Color D raping Q Q e Surface 16 Color D raping Q Q Surface 32 Color D raping Q Q e Q Time Series y1 yn Q e Vector o e e e Specifying M ultiple columns can be specified in a list for example Y1 Y2 Y3 or ina Multiple Data sequence for example Samplel Sample5 Each column must have the same Columns length For example multiple columns can be specified for the Y column for grouped or stacked bar charts area charts or for the Z column for contour or
199. at s New in S Plus 23 N ew Features 23 Welcome to S PLUS Version 4 which integrates the power and functionality of the S PLUS data analysis system and object oriented programming language with a M icrosoft O ffice compatible user interface that is intuitive and fully customizable Because it offers unparalleled power and flexibility S PLUS allows you to create innovative cutting edge analyses N ow all the power of S PLUS is available in an intuitive interface so users of all levels can access advanced statistical methods and revealing graphics In S PLUS data can be imported from virtually any source and can be viewed and edited in the D ata window Point and click control over the details of your graphics makes it easy to produce stunning publication quality output Whether your task is simple or complex S PLUS can lead you to more insightful analysis and new discoveries S PLus is the premier solution for exploratory data analysis and statistical data mining At the core of S PLUS is the S language developed at Lucent Technologies It isthe only language created specifically for data visualization and exploration statistical modeling and programming with data S provides arich object oriented environment designed for interactive data discovery As the exclusive licensee of the S language M athSoft has molded the S technology into the most powerful data analysis product available today T he S PLUS object oriented environment deliv
200. at the second tick mark leave the first row in the column blank and enter your labels starting with the second row D ata Set Specify the name of the data set containing the labels column Prefix Suffix Add text or in text formatting to the beginning or end of each FORMATTING 2D AXES Position Page label In text formatting codes which begin in the prefix and end in the suffix will operate on the entire column of text For instance typing al in the prefix field and in the suffix field causes a column of numbers 1 2 3 to appear on labels as al a2 a3 etc In the Axis Labels dialog the Position page has the following options Overlap O ptions If your tick labels are overlapping you can choose to Skip Stagger or Truncate the labels If you choose Auto S PLUS automatically skips and then truncates unless you have column labels With column labels S PLUS staggers and then truncates Categorical labels however are shrunk staggered then skipped Choose Skip to have tick labels omitted at regular intervals You need to specify the number of tick labels to skip before writing another label in the N umber field For example if you specify 1 every other major tick is labeled If you specify 2 every third major tick is labeled If Auto is specified tick labels are only skipped if they are still overlapping after being reduced by the specified Shrink Factor see the discussion of Shrink Factor in this section
201. ata set will be 210 CREATING TRELLIS GRAPHS used in each panel This is appropriate when all the plots on the graph are using the same data set so that the data columns are all of equal length It is sometimes desirable to combine plots from different data sets on the same Trellis graph You can do this by overriding the conditioning specifications for a single plot For example suppose that you have data from two samples on salt intake blood pressure and age You would like to create a Trellis graph of blood pressure versus salt intake conditional on age for the first sample and then overlay a plot of the second sample conditioned on the same age ranges To create a Trellis graph containing plots from two different samples 1 Turn on conditioning mode by pressing the button on the Standard toolbar 2 Select the x y and conditioning data columns to plot for the first sample and click on the desired plot button A Trellis graph is created 3 Keep conditioning mode on Select the x y and conditioning data columns for the second sample Select the Trellis graph then shift click on the plot button The plot of the second sample will be added Since conditioning mode was on the last column s selected were used to override the conditioning variable s for the second plot The plot specific conditioning specifications are on the plot s Data to Plot page right click on the plot to bring this up Type Select Auto to u
202. ate at the beginning of the session To redo an action w Click on the Redo or Redo List buttons to redo actions For data objects and for script edits clicking on undo for a second time will redo a change To do selective undos using dialog rollback 1 Display the dialog that you used to make the change you wish to undo Fig gt 1 of 2 108 UNDOING ACTIONS To Undo and Redo Using the History Log 2 Click once on the left Rollback button at the bottom of the dialog The fields will change to reflect the previous state of the dialog and the Rollback display will show which of the recorded previous states has been selected 3 ChooseApply or OK You can click on the left Rollback button as many times as desired to go back to previous edits When you have reached the desired edit choose Apply or OK to revert back If you go back too far use the right rollback button to move forward in the H istory log to the desired edit and choose Apply or OK If you only wish to examine previous states you can use the right Rollback button to bring the dialog back to its current settings or simply Cancel the dialog Dialog Rollback just fills in the fields of the dialog for you with previous values It has no effect on an object until you choose Apply or OK Dialog Rollback is closely tied to the History log If you have modified an object in any way visually or through a dialog the corresponding command will appear in the
203. ation of 375 CHAPTER 16 SUMMARIZING AND EXPLORING DATA 376 alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last Im is the most recent linear regression model fit Plot Smooth Check this to display a scatterplot with the kernel smooth Related S PLus language functions ksmooth density loess smooth supsmu CORRELATIONS AND COVARIANCES SPLINE SMOOTHER T his dialog fits a cubic B spline smooth to the input data To perform spline smoothing Choose Statistics gt Smoothing gt Spline Smoother from the main menu The dialog shown below appears aTienthang Spline AE EI ip per Tya Fira isan Degi Fimaks x yabe Dij Pieris ada na oo P mers y r aked Soom pores sired C adap M Umriga Vida bor Enar Fpl EE las Jr iir F Poi bech CE com Hf oon vee Data Data Frame Select or enter the name of a data frame gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame x Variable Select the x variable y Variable Select the y variable Weights Select a column in the data frame to specify weights to be applied to all observations used in the smoothing To weight all rows equally
204. aults are discussed in the following section To set Data window defaults 1 Doubleclick in the upper left hand corner of the D ata window or choose W indow from the Format menu Specify the desired settings in the dialog 2 Right click in the upper left hand corner of the D ata window and choose Save O bject Properties as D efault from the shortcut menu or choose O ptions then Save W indow Size Properties as the D efault 111 CHAPTER 4 USING DATA WINDOWS To Format To format a column Columns 1 Doubleclick in the column header or select the column and choose Selected Object from the Format menu The Column dialog appears Figure 4 11 shows an example 2 Make the desired changes within the Column dialog Choose OK Ooetie Precision Column E _ Parra Format Type Dasma iath h O pi E Justification Fight Deecnplon Figure4 11 The D ouble Pre ison Column dialog Within the Column dialog you can modify the column name description justification display format type width and precision of your data Note You can also right click in a column to display a shortcut menu of options To Change the To change column width by dragging columns Column Width 1 Position the mouse on the line to the right of the column heading for the column you want to change 2 When positioned between the two columns the mouse pointer will become a resize tool 3 Drag the resize tool left to make t
205. axis on the log scale X Axis Style Select the x axis style 519 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS 520 standard scale or extend the maximum survival time creating an axis with extended labels on the right side of the plot T his is the default e tight create an axis labeled internal to the data values extended create an axis whose numeric labels are more extreme than any data values Scale Factor for X Axis Tick Label Enter a number used to multiply the labels on the x axis For example a value of 365 25 will give labels in years instead of the original days Scale Factor for Y Axis Tick Label Enter a number used to multiply the labels on the y axis For example a value of 100 yields a percent scale X Axis Label Enter a label for the x axis Y Axis Label Enter alabel for the y axis Related S PLus language functions for Cox regression coxph print coxph plot coxph predict coxph coxph fit residuals coxph summary coxph survfit coxph Other related S PLus language functions cluster strata Surv survfit PARAMETRIC SURVIVAL PARAMETRIC SURVIVAL This dialog fits a regression moda to survival data See also chapter 22 Overview of Survival Analysis and chapter 25 Parametric Regression in Survival M odels in the Guide to Statistics To perform parametric survival modeling Choose Statistics gt Survival gt Parametric Survival from the main menu The di
206. ayout page has the following options Name Use this option to rename your Graph sheet You can also rename Graph sheets using the Save As option from the File menu Orientation Choose Landscape or Portrait orientation for the page This overrides the page orientation specified under Print Setup Units Specify the unit of measure to be used when creating graph objects on the Graph sheet these are the document units Choose from INCH inches CM centimeters MM millimeters and POINT type setting point size Width Specify the width of the Graph sheet If you choose Printer the width will automatically be set to the printer s paper width with respect to the orientation specified above Height Specify the height of the Graph sheet If you choose Printer the height will automatically be set to the printer s paper width with respect to the orientation specified above Display Margins C hoose whether to display the page margins on the Graph sheet Page margins are represented with dotted lines If the margins are set to 0 no margin lines will appear on the Graph sheet If you choose Printer for your margins you can use the dotted lines as guidelines to make sure you place graph objects within the print area defined by the printer FORMATTING A GRAPH SHEET Top Margin Specify the top page margin for your Graph sheet Choose Printer to have the top margin set to the printer s top margin Bottom Margin Specify the bottom pag
207. base 1 and this data frame has the same 423 CHAPTER 18 FITTING STATISTICAL MODELS 424 Predict page Save number of rows as the number of observations used in the model fit then the saved values are appended to this data frame This allows you to keep predicted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size a warning is issued and a modified name is used Fitted Values Check here to store fitted valuesin the object specified in Save In Working Residuals Check here to store the working residuals in the object specified in Save In The working residuals are the response minus the fitted value Hoalinosr Lowest Squatos Regressice Model Optom Fists Frari Mes Dea A OOS Confeience Level 055 Saree r M Fredis C Gonickorcia loiorsak M Sedan Erom New Data Optional Enter the name of a data frame to use for computing predictions It must contain the same names as the terms in the right side of the formula for the model If omitted the predictions for the original data are computed Save In Enter the name of an S PLus data frame in which predictions confidence intervals and standard errors are saved If an object with the name you enter does not already exist in database 1 then it will be created If you ent
208. bels instead of icons on the tabs A default name is generated if oneis not provided CUSTOMIZNG THE OBJECT BROWSER ToolTip Specify a tooltip for the O bject Browser page When the pointer is held over the tab for this page the tooltip text will appear describing the page A default tooltip is generated based on the class filter if one is not provided Image FileName Enter the complete path and filename for a bitmap file to be used to display an icon on the tab for this page Click the Browse button to navigate to the file A default image is selected if no bitmap file is given Page irio fiimung l eal i Toon a ce a ts Figure6 6 The Page Info tab of the Browser Page properties dialog Using Folders to Organize Work Folders contain shortcuts to objects of all types Through the shortcuts the objects may be modified and used in data analysis Folders may contain shortcuts to data sets functions model objects graph sheets customized dialogs any object that can be viewed in the O bject Browser will admit a shortcut in a folder Folders offer an extremely flexible way to organize work Creating a Folder To create a new folder right click in the left pane of the Object Browser window and select Insert Folder from the context menu or select Insert Folder from the main menu A folder is inserted with its default name in an active edit control Enter a name for the folder or accept the default name If a fo
209. ble test statistics will vary based upon the class of the models being compared Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed The only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names T he default is last anova The saved object is of class anova See the on line help for anova for more information on objects of this type Print Results Check this to print the analysis of variance table for the model comparisons in the designated output window S PLus language functions related to Model Comparisons anova print anova Other related S PLus language functions aov coxph gam gim Im I mRobMM Ime loess nis survreg USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS Multivariate M ultivariate Analysis of Variance 488 Factor Analysis 493 Principal Components Analysis 500 Survival N onparametric Survival 506 Cox Proportional H azards 513 Parametric Survival 521 Time Series ACF Autocovariance Function 526 ARIMA M odading 528 Introduction This chapter describes some dialogs to model multivariate data survival data and time series data T hese include multivariate ANOVA Factor Analysis Cox Regressi
210. bout the model assumptions see the Guide to Statistics To perform fixed effects analysis of variance Choose Statistics gt Analysis of Variance Fixed Effects from the main menu The dialog shown below appears Fined CHocts Ametysis of Warmanoe BE Madel Opiora Pemba Pine Osa Daila Fiame tee hls Subset Rows wih 5e Model Object Fe Orn Pines mat ang akes Dave Aye Piesst nce Poenius Feruk Cae Fimis Data Data Frame Select a data frame gt Tip 444 You can typeinto the D ata Frame edit box any expression which evaluates to a data frame Weights Enter the column that specifies weights to be applied to all observations used FIXED EFFECTS ANALYSIS OF VARIANCE Formula Save Model Object in the analysis of variance To weigh all rows equally leave this blank Subset Rows with Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing V
211. bution of the test statistic to compute the p value Continuity Correction Check this to use a continuity correction in the normal approximation to the distribution of the test statistics Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last lm is the most recent linear regression model fit Print Results Check this to print the results of the W ilcoxon test in the designated output window Related S PLUS language functions wilcox test t test 391 CHAPTER 17 COMPARING SAMPLES TWO SAMPLE KOLMOGOROV SMIRNOV TEST Data T his dialog performs a two sample K olmogorov Smirnov goodness of fit test of two distributions To perform a two sample Kolmogorov Smirnov test Choose Statistics gt Compare Samples gt Two Samples gt Kolmogorov Smirnov GOF from the main menu T he dialog shown below appears Twoeunmpie Kolmogone Seto GoodnessedFil Taal 1 Dera Fou lta fens Hes a Grouping vein Seeds pasties Deta Piene Prt Piewets eed eke yen Dat
212. catter plot on your graph select the scatter plot you want to use Open a view on the data used for the x and y columns in the scatter plot U se W indow Tile Vertical to show your data and graph side by side Click on the Select Point in D ata View button in the top right of the Annotation palette Click on a data point in your scatter plot T he corresponding row in the data view will become selected To select a different row click on a different point in the scatter plot To add a row to the selection shift click on another data point 273 CHAPTER 10 FORMATTING A GRAPH ADDING A CURVE FIT EQUATION If you havea curve fit plot on your graph you can automatically display the equation for the line To insert a curve fit equation 1 Select the desired curve fit plot if there is more than one plot on the graph 2 From the Insert menu choose Curve Fit Equation T he equation of the line is displayed in the Comment dialog T he equation is placed here so you have complete control over its formatting by using in text codes 3 Specify the position for the equation and any desired formatting and choose OK To edit an existing curve fit equation w Doubleclick the equation or select the equation and choose Selected Comment from the Format menu For more information on formatting the equation see section Adding Multi line Text page 261 in this chapter The Curve Fit Equation option is only available when you hav
213. ce Comment plots can be combined with pie charts see the discussion of comment plots earlier in this chapter Plot Margin Start Angle Specify the starting angle in degrees of the first pie slice The default angle is 0 which means the first slice is positioned starting at an imaginary horizontal line drawn from the pie s center to the right edge of the pie The remaining pie slices are positioned after the first slice in a counter clockwise fashion A starting angle of 90 positions the first pie sice at the top of the pie For pie charts you can choose a solid color for each slice fill the pie with a series of shades determined by a range of colors or specify new colors for the filled slices using the standard color dialog See the information on fills in the section Common Plot Properties page 300 for details on fill and color attributes You can have one or two labels for each slice of your pie charts or choose to have no labels at all Label 1 and Label 2 Type For the pie labels you can specify the label style If you choose D ecimal Scientific e Scientific E or M ixed the data values are used as labels If you choose D ecimal Scientific or M ixed the percentage of the pie represented by the slice is used If you select Column you need to specify the column containing the pie labels The column may contain either character or numeric data If you select N one no labels are drawn See the section Formattin
214. choose M enultem from the context menu See the previous sections for details on using the property dialog Menultem objects can be modified with their context menus which are accessible through the O bject Browser The following choices appear after right clicking on aM enultem object in the O bject Browser Insert Menultem Select this to create a new M enultem object Customize T his appears when Action is set to FUNCTION Select this to open the dialog associated to the function Any changes to the dialog persist as dialog defaults Show Dialog On Run This appears when Action is set to FUNCTION Check this to cause the dialog associated to the function to open when the item is selected See Table 22 1 for details Always Use Defaults This appears when Action is set to FUNCTION Check this to force the use of the default values when the function executes See Table 22 1 for details S PLUS makes a distinction between the default argument values for a function as defined in the function s dialog via the Function nfo object and as defined by the function itself Always Use D efaults refers to the dialog defaults Hide Select this to hide the menu item It will not appear in the menu system and the M enultem object icon will appear grayed out Delete Select this to delete the M enultem object The menu item will no longer be available 589 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE Displaying Menu Items Manipulating Menu I
215. choose the Select All command or 1 Position the mouse pointer in one corner of the Graph sheet 2 Click and drag until the selection box encompasses all objects on the Graph sheet All objects are selected and have green selection knobs displayed on their outline To select all objects of the same type A number of objects of specific types can be selected using the Edit menu 1 From the Edit menu choose the Select command A list of available object types will appear 2 Select from the list for example All Plots All plots on the Graph sheet will be selected 283 CHAPTER 11 WORKING WITH GRAPH OBJECTS SUMMARY OF FORMAT AND VIEW MENUS Selected Objects Ctrl 1 Change Plot Type Sheet 4oply Style Align Distribute Snap to Grid Plot Summary rrange Graphs Exchange Graphs Ering to Front Send to Back Bring Forward send Backward Figure11 1 The Format menu when a Graph sheet is active Table 11 1 The Format menu options Option Purpose Font Set the font when editing text Selected O bjects M odify the properties of all the selected objects Change plot types Convert from one plot type to another for example scatter to bar 284 SUMMARY OF FORMAT AND VIEW MENUS Table 11 1 The Format menu options Plot Summary Arrange Graphs Exchange Graphs Bring to Front Send to Back Bring Forward Send Backward Option Purpose Sheet M odify the proper
216. cified colors Choose Special Colors to specify customized colors in the Special Colors dialog Choose Color Numbers Column to specify a column of color numbers in a D ata window Choose Color Numbers Column to specify a column of colors in a data set For example if you have three color numbers specified in your color column for example 1 5 7 the surface will be drawn in the three different colors according to height T he lowest third will be drawn in the first color blue the next highest third will be drawn in the second color magenta and the highest third will be drawn in the third color white The IBM color numbers range from 0 to 15 You can also specify 16 additional custom colors and shades between 5 colors 4 intervals for a total of 64 colors To specify custom colors choose Sheet from the Format menu click on the Color page and click on the Edit Colors button Custom colors specified here will appear in all color lists for the current Graph sheet To save the custom colors for use with all subsequent Graph sheets you can save the Graph sheet defaults Right click on the Graph sheet and choose Save Graph Sheet Properties as D efault from the shortcut menu Fill Color For a solid fill type specify a single color for filling the 3D surface mesh D ata to Vary Colors By T he colors in the surface are varied by the column of data specified here Choose to vary the colors by the X Y Z or W data columns or specify O the
217. ck this to save the residuals from the final additive fit Pearson Residuals Check this to save the pearson residuals T hese are standardized residuals on the scale of the response Deviance Residuals Check this to save the deviance residuals The sum of squares of these add up to the deviance Response Residuals Check this to save the response residuals T hese are the response minus the fitted value Geeealired Linowa Models Mode Opin re Fiera ae Fiz Pinte Pans Aandal Plots F Amida va Ft P Plot Partie Apsicuests DO Seger Aonta sa Fi E F Resporesws Fi p E Rondisi Homa g F Opsons M ideda Smooth M aciada Rugpiot Humb of Earme Foire Te ldeathy 1 461 CHAPTER 18 FITTING STATISTICAL MODELS 462 Plot Page Plots Options Partial Residual Plots Residuals vs Fit Check this to plot the deviance residuals versus the fitted values Sqrt Abs Residuals vs Fit Check this to plot the square root of the absolute values of the deviance residuals versus the fitted values This plot is useful for checking for the constant variance assumption of the model Response vs Fit Check this to plot the response variable versus the fitted values The line y x is also drawn on the graph Residuals Normal QQ Check this to create a Normal quantile quantile plot of the pearson residuals Include Smooth Check this to display a smooth curve computed with loess smooth on the Residuals vs Fit Sqrt Abs Residuals vs Fit
218. close the transformation dialog Click on the button Response as in Step 3 to add log Fuel as the response to the formula Similarly you may use a transformed variable from the Choose Variable s list box as an explanatory variable Accept the formula To pass the formula back to the Linear Regression dialog you can either click on the OK or Apply button If you click on the OK button the formula dialog is closed and the formula Fuel Weight M ileage is shown in the Formula of the Linear Regression dialog You can now continue the Linear Regression dialog If you click on the Apply button the formula Fuel Weight M ileage is passed to the Linear Regression dialog while keeping the Formula dialog open In this case you cannot edit the D ata Frame and the Formula group in the M odel page of the Linear Regression dialog Similar Formula Builders The following statistical models use the same Formula Builder as Linear Regression Logistic Regression Log Linear Regression Robust Regression Generalized Linear M odels and Generalized Additive M odels The following use the Linear Regression Formula Builder with minor differences Fixed Effects ANOVA Random Effects ANOVA MANOVA Tree models Local Regression and Stepwise Linear Regression The following use a much simplified version Crosstabulations Factor Analysis and Principal Components 540 COX PROPORTIONAL HAZARDS COX PROPORTIONAL HAZARDS Use the Creat
219. conditioned plot shows a clear positive relationship between the oxides of nitrogen and the compression ratio for low values of E For high values of E the slope is close to zero In each panel the pattern appears linear Changing the plot type Since the patterns appear to be linear let s redraw the Trellis graph using a least squares fit 8 Click on any symbol in the plot to select it N ow click on the Linear Fit plot button on the 2D plot palette The graph will be redrawn 217 CHAPTER 9 WORKING WITH TRELLIS GRAPHICS using a linear fit Figure9 5 Examining pollutantsin automobile exhaug Example 2 Barley Yields By Site and Year 218 This example shows a Trellis display of data from an agricultural field trial to study the crop barley At six sites in M innesota ten varieties of barley were grown in each of two years The data are the yields for all combinations of site variety and year so there are 6 x 10 x 2 120 observations T he barley experiment was run in the 1930 s T he data first appeared in a 1934 report published by the experimenters Since then the data have been analyzed and re analyzed R A Fisher presented the data for five of the sites in his classic book The Design of Experiments Publication in the book made the data famous and many others subsequently analyzed them Then in the early 1990 s the data were visualized by Trellis Graphics The result was a big surprise Throug
220. contiguous columns 5 Position the mouse within the selected region not in the column header until the cursor changes to an arrow Pressing the left mouse button drag the data from the D ata window or O bject Browser and move it over the plot icon W hen the plot icon changes color release the mouse button to drop the data and generate the plot For information on adding axes see Chapter 10 Formatting a Graph Adding a Plot To add a plot using the Insert Plot option Using the Menus 1 If you have more than one graph on the Graph sheet select the graph you want to add a plot to by clicking outside the axes or on the graph border 185 CHAPTER 8 CREATING A GRAPH Hiding Unhiding and Deleting a Plot 186 2 From theInsert menu select Plot 3 Select the desired plot from the Plot Type list Choose OK to close the dialog and continue T he plot property dialog appears 4 Specify the columns of data to plot on the D ata to Plot page of the dialog 5 Makeany other desired changes to the plot specifications and choose OK You may want to hide a plot temporarily instead of deleting it from the graph You can hide the plot and still retain all of its specifications T his can be useful if you want to edit parts of a graph containing a complicated plot type that takes a long time to redraw To hide a plot 1 Doubleclick the plot or select the plot and choose Selected Plot from the Format menu or right click on
221. ct Browser D ata W indow Graph Sheet Commands Window Script W indow Report Window S PLUS M enus S PLus D ialogs S PLus Palettes and Toolbars Chapter 3 Tutorial Introduction Quick Tour In D epth Tour T he O bject Browser Inserting a N ew Column Creatinga2D Graph Creating 3D Graphs Using Trelis Graphics for M ultipane Conditioning Applying Statistics M odels T he Commands W indow Creating PowerPoint Slides Automatically Chapter 4 Using Data Windows Working with D ata W indows Creating a N ew Data Window D ata O bjects Editing a D ata Set The Current D ata Window Selecting Cells Columns and Rows Selecting Cells Extending the Cell Selection Selecting Columns Selecting Rows U sing Keyboard and M ouse Shortcuts Using theGo To Cell Option Column N ames and Column N umbers Column Listsin the D ata W indow Entering D ata Entering D ata from the K eyboard Entering D ata from O ther Sources Editing Data Editing T ime Series D ata M oving and Copying D ata Inserting and D eleting Cells Rows and Columns Inserting Columns and Rows D eleting Data Sorting D ata Undoing Actions Formatting D ata Windows To Format Columns Chapter 5 Importing and Exporting Data Importing D ata Files The D ata Import Dialogs CONTENTS Filter Page Notes on Importing Files N otes on Importing ASCII Delimited ASCII Files N otes on Importing Excel Files N otes on Importing Files with M ultiple T ables N otes on Importing Lotus Fi
222. ct will be saved as alist with components named x and y Plot Loess Curve Check this to generate a scatterplot with a corresponding loess curve Related S PLus language functions loess loess smooth panel smooth scatter smooth 371 CHAPTER 16 SUMMARIZING AND EXPLORING DATA SUPERSMOOTHER This dialog produces a smooth curve through the input data using a nonlinear variable span smoother To perform scatterplot smoothing using the supersmoother Choose Statistics gt Smoothing gt Supersmoother from the main menu The dialog shown below appears Supersmoother OE xi m Data r Options Data Frame ethanol IM Use Cross Validation to Set Span x Variable Smoothing Span fos I x Variable is periodic Bass Frequency jo y Variable Results Weights Tr H Save s flast supsmu Plot Smooth Cancel Apply current Help Data Data Frame Select or enter the name of a data frame gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame 372 x Variable Select the x variable x Variable is periodic Check this if the values of x Variable are in the range 0 0 1 0 and havea period of 1 y Variable Select the y variable Weights Select a column in the data frame to specify weights to be applied to all CORRELATIONS AND COVARIANCES Options Results observations used in the smoothing To we
223. ction we will fit linear regression models to predict Ozone using Temperature Radiation and Wind 77 CHAPTER 3 TUTORIAL Summaries First we will look at summaries of the data in the Exenvirn data frame 1 Use the Find button on the Object Browser to find the Exenvirn data frame 2 Select Statistics D ata Summaries Summary Statistics 3 In the Data field type Exenvirn or alternatively use the drop down list Click OK 4 Summaries for the columns will appear 5 Click on the O bject Browser 6 Click on the data frame icon 7 Select Exenvirn 8 Select Statistics D ata Summaries C orrelations 9 The Correlations and C ovariances dialog will appear with Exenvirn selected as the data frame Click OK Correlations for the columns will appear in the Report window Linear model N ext we will use the Linear Regression dialog to fit alinear model predicting O zone from the other variables Simple model from a dialog 1 Select Statistics R egression Linear T he Linear Regression dialog will open 2 In the Data Frame field type environmental or select it from the drop down list 3 In the formula field type ozone radiation temperaturetwind or press Create Formula to launch the Formula Builder dialog The Formula Builder lets you describe complex regression models by selecting variables and indicating how they are used in the modal To use the formula builder Select O zone in the Choose Variable s list C
224. ctly The elements of irregular time series created with the i ts function can be edited in the same way as row names by double clicking on the cell in the time series column and entering new data 99 CHAPTER 4 USING DATA WINDOWS MOVING AND COPYING DATA You can move and copy data cells within a data set across data sets and into Graph sheets See the chapter Creating a Graph for information on adding data to graphs See section Selecting Cells Columns and Rows page 90 for more information on selecting itemsin a D ata window In S PLUS there are three ways to move and copy data cells in D ata windows 1 2 3 D ragging selected data cells Using the Cut Copy and Paste options Using the D ata menu options To Move or Copy To move or copy cells by dragging Cells 100 1 3 Select the cells you want to move or copy Use cTRL click to select discontiguous columns or rows Position the mouse pointer within the selected cells The mouse pointer changes to an arrow To move cells hold down the left mouse button and drag the selected cells to the new location Release the left mouse button to move the selected cells to the new location paste area To copy the cells hold down the CTRL key while releasing the mouse button This will work when copying or moving between D ata windows as long as you arrange your D ata windows so you can see both the source and target cell locations To move or copy c
225. current settings in Print Setup To print using the Print Dialog 1 From the File menu choose Print 2 In the Print dialog choose the options that you want 3 ChooseOK to start printing To cancel printing click on the Cancel button or press ESC Graph Sheet Print to File Choose this option to have your Graph sheet printed to a file Printing Options instead of directly to a printer When you accept the Print dialog you will be prompted to specify a file name Data Window Margins Specify the margins for the data window on the output page Printing Options Print Column H eadings Choose whether to have column headings printed on each page of your D ata window Column headings are printed at the top of each page Print Row H eadings C hoose whether to have row headings printed on each page of your D ata window on the left side Print Grid Lines Choose whether to print grid lines on each page of your D ata window Sending System Requirements Electronic Mail To use document mailing you need one of the following e Microsoft Exchange or other mail system compatible with the M essaging Application Programming Interface M API e Lotus cc Mail or other mail system compatible with Vendor 338 EXPORTING GRAPH SHEETS TO DIFFERENT FILE FORMATS Independent M essaging VIM Note Document mailing may not work across electronic mail gateways To Mail a Document 1 From the File menu choose Send 2 C
226. d You can selectively undo or redo commands for a selected object To do this select the object before choosing H istory D isplay Then in the Display dialog choose the Selected O bject option and choose OK The History log displays only the commands for the selected object If you want to redo a portion of a script you can select the desired commands before running the script O nly the selected commands will be executed For more information on using the History log see section Working with Scripts page 547 FORMATTING DATA WINDOWS FORMATTING DATA WINDOWS Data Window Properties To Set Data Window Defaults To format a Data window 1 Doubleclick in the top cell in the upper left hand corner of the D ata window 2 Make any desired changes and choose OK Name Specify the desired name for the Data window This option is a convenient way to rename your D ata objects Default Column Type Specify the default column type for all empty columns in the Data window This option will not change the column type of existing columns Use the Change D ata Type button on the D ata toolbar to change existing columns Font Size C olor C hoose a font type size and color for the D ata window Bold Italics C hoose if you want the D ata window font to be bold or italic You can set Data window defaults name row height and default column type You can also set column defaults justification width and precision Column def
227. d F Hnc A E cel Smeeivets Sin Hoala Object OOOO D Egars Ha ot Par Formula Fron Crosia Famila Model Page Data Data Frame Select a data frame gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame Weights Enter the column that specifies weights to be applied to all observations used 433 CHAPTER 18 FITTING STATISTICAL MODELS Local Smoothness 434 Model in the local regression To weigh all rows equally leave this blank Subset Rows with Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the model If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Normalize Numeric Pred
228. d Rows Deleting Data Sorting D ata Undoing Actions 106 108 85 CHAPTER 4 USING DATA WINDOWS Working with Data Windows 86 Formatting D ata Windows 111 To Format Columns 112 S PLUS lets you edit your data sets as columns of information that can be displayed in D ata windows You can have many different D ata windows each displaying a different data set You can refer to columns in separate data sets by specifying the name of the data set along with the column name separated by a dollar sign for example test1 x An S PLUS D ata window is similar to a spreadsheet but is column oriented rather than cell oriented D ata windows provide access to powerful features for editing and transforming data T hese features include e Column row and block operations as well as individual cell operations Both simple and advanced statistical analysis D ata sets can contain more information than can be viewed on the screen at one time D ata windows allow you to look at portions of data Simply scroll to the parts of the data you want to view using the scroll bars or cursor keys or use the Go To Cell option to move quickly to any row or column in the data Data windows can be duplicated to allow concurrent views of many different sections of the data You can have as many Data windows as you want The Window menu displays a partial list of all data sets currently in windows If you have more than nine windows open you can
229. d data shown in each panel and set the aspect ratio of each panel to 0 5 5 Right click in a blank area inside of the graph to get the shortcut menu for the graph properties Choose M ultipanel 6 Under Conditioning Columns set the Order Type to M edian of X Under Layout set the of Columns to 1 Click on the Position Size tab of the dialog Set the Aspect Ratio to 0 5 Click OK Adding a legend and editing the plot N ow add some final touches to your plot add a legend and vary the symbol styles as well as the symbol colors for the two years 7 Click on the legend button of the graph toolbar A legend will be automatically created and placed on your graph Select the legend by clicking just inside the border of the legend Drag it to the desired location 8 Right click on any symbol on your plot to access its short cut menu Choose Vary Symbols On the Vary Symbols page set Vary Style By z Column so that both the color and symbol styles will vary by year If you would like to change the symbols and colors used they are specified under O ptions Graph Styles C olor Press O K N otice that 219 CHAPTER 9 WORKING WITH TRELLIS GRAPHICS 220 the legend has been updated to the new symbol styles N ow examine your graph to find the undetected happening It appears in the Morris panel For all other sites 1931 produced a significantly higher overall yield than 1932 The reverse is true at Morris But most importantly the
230. d disk space required 40M B Typical installation 108M B Full installation M icrosoft Windows 95 WindowsNT or Windows 3 1x VGA Super VGA or most other Windows compatible graphics cards and monitors OneCD ROM drive local or networked M icrosoft M ouse or other W indows compatible pointing device W indows compatible printers are supported HELP SUPPORT AND LEARNING RESOURCES HELP SUPPORT AND LEARNING RESOURCES Getting Help T here are a variety of ways to accelerate your progress with S P LUS and to build upon the work of others This section describes the learning and support resources available to S PLUS users Online Help S PLUS offers an online help system to make learning and using S PLUS easier Under the H elp menu you will find options for Using S PLu s how to use the graphical user interface Language Reference details on each function in the S PLUS language Questions and Answers some common difficulties and proposed solutions Online M anuals see below and Visual Demonstrations T here is also context sensitive help accessed by clicking on the H elp buttons in the various dialogs or by clicking on the context sensitive H elp button on the toolbars There is also Language Reference help available through the S PLUS Commands window by typing hel p at the S PLUS prompt or by pressing the F1 key while S PLUS is active Printed and TheS PLus Programme s Guide the Guide to Statistics and the S
231. data are automatically interpreted in document units with each comment s position measured from the lower left hand corner of the Graph sheet X Axis Y Axis Choose which x axis and y axis to scale the comment plot to By default the plot is scaled to the first x axis and y axis on the graph Horizontal Justification Specify the horizontal justification for your comment text in relation to the x y position left right center corner For example selecting left places the comment text to the left of the data point T he default horizontal justification is center Vertical Justification Specify the vertical justification for your comment text in relation to the x y position up down center corner For example selecting up places the text above the data point The default vertical justification is center H orizontal Vertical O ffset Specify the distance in inches cm between the comment text and the position defined by x and y Format If your Z column is numeric or blank specify the display format decimal scientific mixed or auto Precision Specify the number of digits to display after the decimal For plot options you can specify standard fill attributes color fill and background and border attributes style color and weight for the box drawn around the comments Additional options include H orizontal Vertical Margin Specify the vertical and horizontal distance between the comment text and the box around the text
232. database 1 then it will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame T his allows you to keep predicted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Predictions Check this to save the predictions in the data frame specified in Save In Confidence Intervals Check this to store lower and upper confidence limits in the object specified in Save In The column names will be N L C L and N U C L where N is 100 times the value specified in Confidence Level These confidence limits for the mean response are computed as the prediction plus or minus t 419 CHAPTER 18 FITTING STATISTICAL MODELS 420 Options value times standard error Standard Errors Check this to store the pointwise standard errors for the predictions in the object specified in Save In Confidence Level Enter the confidence level to use when computing confidence intervals T his value should be less than 1 and greater than 0 S PLus language functions related to Linear Models im plot Im predict im print Im
233. der If no folder is specified the toolbar will contain a single button with the ToolbarButton defaults Document Type Select the document types which will when in focus allow the toolbar to be visible Click OK and anew Toolbar object appears in the O bject Browser Creating and To add a button to an existing toolbar right click on the corresponding Modifying Toolbar object in the Object Browser and select New Button from the context menu T he ToolbarButton property dialog appears asin figure 22 2 Buttons Todibaruton Object 31 Ail i Baton Commend image Diarra Peroni Tip Teer I I ETTO F Hda Arton NOME F Detetabis Figure 22 2 The ToolbarButton property dialog Name Enter aname for the new button Type Select BUTTON to create a button or select SEPARATOR to create a gap between buttons in the toolbar Action This applies to ToolbarButton objects of type BUTTON e NONE No action is performed when the button is clicked 569 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE 570 BUILTIN One of the actions associated with the default menus or toolbars is performed when the item is selected T hese are listed on the Command page in the Built In Operation drop down box This option allows you to use in a customized toolbar any of the intrinsic menu or toolbar actions such as W indow C ascade e FUNCTION Under this option an S PLUS function is executed when the button is clicked Optionally the di
234. directly Print Digits T he number of significant digits to use in print and therefore in automatic printing Setting this to 17 will give the full length of double precision numbers Time Series Eps Specify the time series comparison tolerance This small number is used throughout the time series functions for comparison of their frequencies Frequencies are considered equal if they differ in absolute value by less than the number specified here Editor Specify the default text editor command used by the edit function W hatever editor you choose will be invoked in the style of N otepad that is 599 CHAPTER 23 CUSTOMIZNG YOUR S PLUS SESSION Specifying Command Window Settings Font Options Page Specifying Undo amp History Options 600 by a command of the form Notepad filename followed by the reading of editing commands Do not supply editors that expect a different invocation or a different form of user interaction Pager Specify the default pager program to be used in the hel p and page functions Whatever pager you choose will be invoked aS pager filename and should read from filename To specify the Commands window settings choose Commands window under the O ptions menu T he Command window settings dialog will appear Specify the font font size color and styles used in the Commands window In order to have proper alignment of output we recommend a fixed width font such as Letter Gothic or Courier
235. dow history Number of conditioning variables Turn on context help S PLUS PALETTES AND TOOLBARS Graph Toolbar Table 2 4 The Graph toolbar see graphics section arting with chapter 8 Creating a Graph Graph z 7 Set font Arial ii Set text size 12 ha Bold text Italic text Underline Superscript Subscript HZ7 U Send to Bring to mal Tal back front Fill color Line Pattern Line style Line weight Symbol color E g Auto legend Annotation Send graph palette to other application 49 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE Data Frame Toolbar Table 2 5 T he D ataFrame toolbar see chapter 4 Using D ata windows Align left Increase precision Insert col Remove column Sort ascending Center Align right D ecrease Change data type precision Column type to insert Clear Remove Clear row column row Sort Adjust Increase D ecrease descending column column column width to fit width by with by one one Script Toolbar Table 2 6 The Script toolbar see chapter 21 Working with Script Commands and Report windows script Run selected script Find text 50 S PLUS PALETTES AND TOOLBARS Object Browser Toolbar Table 2 7 The O bject Browser toolbar see chapter 6 Using the O bject Browse Create O bject Expand Collapse Find Browser page item item S PLUS objects Report Window Toolbar Table 2 8 T he
236. down the left button move the cursor until the desired size and shape brush is drawn then release the button Big Points Click on the box marked Big Points to use a larger plotting symbol To return to smaller symbols click on this option again Labels Click on the box to have points within the brush area labeled with their row label When you move the brush with any mouse button down labels for points no longer within the brush area will disappear If you move the brush with no button pressed labels will persist until you press any button 205 CHAPTER 8 CREATING A GRAPH Spinning Data 206 Persistence By default Persistence is on Click on the box to turn it off W ith the option off your next highlighting operation highlights only those points within the brush and downlights all previously highlighted points that are not currently within the brush If you selected the Spin W indow option you should have a Spin sub window in the upper right of your screen It contains a cloud of points with three axes labeled with the names of the first three variables Using the Spin Buttons U se the five buttons marked with arrows to spin the data cloud When you press and hold the left mouse button on any of the direction buttons the data will spin in the appropriate direction Reset Click on the Reset button to reset the axes and the point cloud to their original positions Speed Use the horizontal scroll bar marked Speed to adjust
237. duals Check this to save the residuals Save In Enter the name of an S PLUS data frame in which fitted values and residuals are to be saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that 453 CHAPTER 18 FITTING STATISTICAL MODELS 454 Plot page Plots already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame T his allows you to keep fitted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Aaedom fegis Anahyeis of Varaace aE Model Optra Pesta Pint Plots Opbons P Amida wa Ft P cude Emoni I Sep Abs Ronekisa wa Fa T inchuide Reg pia VD Response wn Fi Hunia of Eiramaa Pons To kienty P Ronida Homa On Purial Rasidial Pols P Plot Pate Risickuels l AeiiduetFi Soraa l Cooks isare x r Residuals vs Fit Check this to display a plot of the residuals versus the fitted values Sqrt Abs Residuals vs Fit Check this to display a plot of the square root of the absolute values of the residuals versus the fitted values This plot is useful for checking f
238. e Draw Conf Bounds f enabled confidence intervals are shown If the confidence intervals on two boxes do not overlap this indicates a difference in location at arough 5 significance level Notched Boxes If enabled confidence intervals will be notched Comment plots can be created using the Graph menu on the Standard toolbar Specify the names or column numbers of your X Y and Z columns The X column specifies the X position of each comment the Y column the Y position of each comment and the Z column contains the comment text If no Z column is specified the X Y coordinates are displayed on the graph for example 10 2 Table 12 3 Sample data for a comment plot X Y TEXT 1 0 1 62 M onday 2 0 2 65 Tuesday 3 0 1 98 Wednesday 4 0 4 52 Thursday 5 0 1 76 Friday Use Axes Units Choose to position the comments relative to an axes pair on the graph 2D and Polar axes only If you choose this option your x and y data values are interpreted just like x and y data points on the graph Otherwise your x and y data values are interpreted in document units 309 CHAPTER 12 EDITING PLOT PROPERTIES Position Box Font styles 310 instead of axes units For example an x y location of 1 2 positions the associated comment one inch up from the bottom of the sheet and two inches from the left side of the Graph sheet If you specify 3D axes types or have no axes this option is not available Your x and y
239. e Explode in cm Specify the amount in inches cm of offset from the center for each pie slice T he default is 0 for no explosion Polar plots can be created using the Graph menu on the Standard toolbar Polar plots display data in polar coordinates radii lengths and angles You can create a polar plot by choosing either Polar Scatter Plot or Polar Line Plot and specifying the names or column numbers of your X and Y columns TheX and Y columns must be the same length the X column should contain the radii and the Y column should contain the angles in radians If your Y column data isin degrees you can select Convert from the D ata menu to use the degtorad function to translate the data into radians For polar lines you can specify standard line attributes style color and weight T he line style is set to N one for polar scatter plots You can also choose to break lines at missing values or at symbols For polar line and scatter plots you can specify standard symbol attributes style color height line weight and frequency Use Text as Symbols Specify a string of text in any font to be used as a symbol for a line scatter plot Symbol Text Specify the text to be used as a symbol PLOT PROPERTIES Quantile Quantile Plot Data to Plot Vector Plots Data to Plot Arrow Vector Options Font Choose a font type for the text Bold Italics Underline Specify text formatting options See section Formatting Polar
240. e Command line call the guiSave Report function For example guiSave Report Name Report FileName C temp report2 rtf will save the default report in the filec temp report2 rtf To create a new Report window 1 From the File Menu choose New A scroll box will open Select Report file 2 From the command line call the guiCreate Report function For example guiCreate Report Name ShortReport will create anew Report window called ShortReport T he function call menuDescribe data FuelFrame min p T first quant p T mean p T median p T third quant p T max p T nobs p T valid n p T var p F stdev p T sump F factors too p T print p T will populate the newly opened Report window with some text from the data frame Fuel Frame How to Open a file in the Report window 1 File M enu choose O pen and select a file with an rtf extension 2 Command line gui Open Report For example to open a text filein a Report window gui Open Report FileName C temp reporti txt To open an RTF filereport2 rtf in aReport window gui Open Report FileName C temp reporti rtf How to set preferences for the Report window 1 From the Options M enu Select Text output window to display a dialog box with radio buttons Select the Report button 564 THE REPORT WINDOW 2 From the Command line Call the guiSet Output WindowPreference function For exa
241. e File menu in S PLUS You will see the Welcome screen of the PowerPoint Presentation W izard Click N ext Several options are available on this page You can Click the Add Graph button to find S PLUS graphs on your system that you would like to add to the list for this presentation Click the Load List button to load a previously saved presentation list of graphs Click Save List to save the current presentation list Click Clear List to reset the contents of the current presentation list To rearrange items in the list select a graph in the list and click the Up or Down button to move it up or down through thelist To remove a graph from the list select it and click on the Remove Graph button 4 Click Next to move to the next page of the wizard 5 Click Finish PowerPoint is started and the graphs you chose are inserted in slides in a new PowerPoint presentation in the order you specified in the presentation list in the wizard While the graphs are inserted status information appears in a box in the wizard that tells you what is going on After the presentation is complete it will be automatically saved to disk with the name you gave the presentation list in the wizard CREATING A POWERPOINT PRESENTATION If you did not save the presentation list in the wizard and the Presentation Nameis Untitled then PowerPoint will not save the presentation when the wizard is finished preparing it You will need to e
242. e Formula button from the M odel page of the C ox Proportional H azards dialog to open the Formula page Formula ix m Variables m Survival Response Choose Yariable s Nos E Time 1 4 Zensoring Indicate Transformation 23 TR Term s Main Effect Interaction ms Main nteract i Cluster HI Strata Offset Add Response r E emove Term Term Remove Formula OK Cancel H current Figure 20 3 Cox Proportional H azards dialog Formula page Tip A data frameisrequired to open the formula page W hen the formula pageisup theD ata Frame Type of Censoring and Formula group in the moda page are disabled until theOK or Canca button in the formula page is clicked If a formula is composed and the Apply button is clicked the formula is sent back to the modd page You may repeatedly enter formulas and apply them to the model page A formula can be autogenerated using the variable list in conjunction with the available buttons T he Formula field can also be edited directly Variables Variables belonging to the Data Frame selected in the model page are displayed in the Choose Variable s multiple selection box Use the CTRL key to select multiple variables if appropriate T he variables list is used to define the survival response and model predictors To perform transformations on selected variables click on the Transformation button 541 CHA
243. e Insert menu select Titles 2 From the Titles submenu choose M ain or Subtitle S PLUS opens an edit box for you to enter text 3 Type the desired text Press ENTER to create a new line To end editing and save results click outside the edit box Alternatively you Can press F10 Or press CTRL ENTER To exit without saving press the ESC key You can either type in some text or choose from one of several automatic titling options If an automatic titling option is entered text is taken from the description name or column number assigned in the Data window These are automatically updated when changes are made to the Data window Auto which is the default may be specified in the title text field Aut o cannot be combined with any other characters within a title field For example if you type AutoXYZ the XYZ text is ignored The sign is ignored if it isnot the first character typed within a title field For example if you type Aut o the word Auto is displayed and is skewed Table 10 2 Automatic titling options Option Description Auto S PLUS uses the X Y or Z column description if one was specified If no column description is found S PLUS uses the column name if one was specified If neither a description or name is found S PLUS uses the data set name followed by the column number T he only exception to this isin the case of grouped bar charts with multiple columns of Y data when by default the
244. e Page tabs on the bottom of the graphs to sdect different pages 80 IN DEPTH TOUR TheCommands for some analyses it is more convenient to work with an interactive data Window analysis language than to maneuver through a series of dialogs We will use the Commands window to fit another linear model and perform some diagnostics If it is not already open open the Commands window using the Commands window button on the Standard toolbar Figure 3 26 The Commands window button on the Standard toolbar Figure3 27 TheCommandswindow gt prompt and vertical bar cursor The Commands window uses a gt prompt In this document text starting with gt isto be typed at this prompt do not type the gt Listing S PLUS 1 To list S PLUS objects available in the working directory type Objects gt objects 2 Recall that the default option in our linear model dialog was to save the model as ast 1m Note that such an object does indeed exist To see the brief summary for the modal type gt last Im 81 CHAPTER 3 TUTORIAL Fitting a Linear Model Creating Plots from the Commands Window Editing a Graph 82 We noticed in the Trellis graphic that there appears to be an interaction between temperature and radiation in determining ozone level We will fit a model containing interactions and explore whether the interactions are significant 1 To fit a linear model with all two way
245. e Row prompt it does not have to be the first row You can select arectangular subset of your worksheet by specifying starting and ending columns and rows Lotus style column names eg A AB can be used to specify the starting and ending columns The row specified as the starting row is always read first to find out the data types of the columns Therefore there cannot be any blank cells in this row In other rows blank cells are filled in with missing values 127 CHAPTER 5 IMPORTING AND EXPORTING DATA Notes on Importing dBase Files Notes on Importing and Exporting Access Files Notes on Importing FASCII Formatted ASCII Files 128 S PLUS imports dBase and dBase compatible files The file name and file type are often the only things you need specify for dBase type files Column names and data types are obtained from the dBase file H owever you can select a rectangular subset of your data by specifying starting and ending columns and rows All imports from and exports to Access are done using ODBC so the various OD BC components must be properly installed You can use FASCII import to specify how each character in your imported file should be treated For example you must use FASCII for fixed width columns not separated by delimiters if the rows in your file are not separated by line feeds or if your file splits each row of data into two or more lines For FASCII import you need to specify the file name and
246. e at least one curve fit plot on the graph If you have multiple curve fit plots you can select each one and get the equation describing the line automatically 274 ADDING LINES SHAPES AND SYMBOLS ADDING LINES SHAPES AND SYMBOLS You can add extra items to your graph such as text lines shapes and symbols These drawing objects can be added by using the Annotation palette or by using the Annotation option on the Insert menu A short description of each drawing object is shown later in this section Adding You can add drawing objects to a graph by using the Annotation option on Annotations the Insert menu T his allows you to specify both the property attributes of an Using the Menus object and its position on the graph within the same dialog To add an annotation using the Insert Annotation option 1 From the Insert menu choose Annotation A submenu appears listing the drawing objects 2 Choose the object you wish to draw on the graph The property dialog for the object will appear Specify the desired properties including the x y position for the object 3 ChooseOK The object is drawn on the Graph sheet Adding You must have the Graph toolbar displayed to access the Annotation palette ee You can add a drawing object to a plot in two different ways sin e g 1 Drag and drop a drawing object icon from the Annotation palette Annotation Palette onto the graph 2 Click adrawing object button to turn the mouse into a dra
247. e desired model Create Formula Click this to open a formula builder dialog used to construct a formula specifying the desired model See the chapter Building Formulas for more information Local Reg isints Model Opto Firgubs Pint Corm Faama Frisos Fitieg i s T imapa Tuil Biew he tis _ Surface Fitting Choose direct to use observed responses directly in the surface fitting By default interpolated points are used in the fitting Cell Size 435 CHAPTER 18 FITTING STATISTICAL MODELS Enter the number of cells used in locally fitting This field is enabled only when interpolation is chosen above No of Iterations Enter the number of iterations This field is enabled only when Family is specified as symmetric on the model page gt Tip Most users don t need to s the control parameters since the default provides very satisfactory performance in most of the cases Changing these parameters can substantially burden the computation for large data sets Local R griiiiii Moda Options Biegu Pios FF Prrd Apnd Saad Rasia Seve in T ited eles P Residuals Results Page Saved Results Print Results Check this to print the results of the fitted local regression in the designated output window Save In Enter the name of an S PLUS data frame in which fitted values and residuals are to be saved If an object with the name you enter does not already exist in database 1 then i
248. e distance between the bottom of each line of legend information asa proportion of the text height For example a value of 2 produces a vertical space that is twice the text height Middle Spacing Specify the distance between the sample and the associated legend text in document units Vertical and Horizontal Margin Specify the distance between the legend contents and the legend box in document units Round Corners C hoose whether to have rounded box corners Horizontal and Vertical Radius Specify the degree of rounding for the box corners The larger the value specified here the more rounded the corners appear In S PLUS you have complete control over the formatting of each item in the legend To edit a legend item w Doubleclick the legend item or select the legend item and choose Selected O bject from the Format menu In the Legend Item dialog you can specify the item text and the formatting attributes of the legend sample 271 CHAPTER 10 FORMATTING A GRAPH Text Page Line Symbol Page Fill Border Page Font Page 272 In the Legend Item dialog the Text page has the following options Text Enter the text description for each legend item Auto is the default for the legend text fields The Auto specification uses the column name or the column number from the data column s specified for Y for most 2D plots and Z for most 3D plots A separate legend item is created for each bar in a standard bar chart each gr
249. e margin for your Graph sheet Choose Printer to have the bottom margin set to the printer s bottom margin Left Margin Specify the left page margin for your Graph sheet Choose Printer to have the left margin set to the printer s left margin Right Margin Specify the right page margin for your Graph sheet Choose Printer to have the right margin set to the printer s right margin In the Graph Sheet dialog you can also specify the default positioning of your graphs The following offset values are measured from the lower left hand margin setting Top Offset Specify the distance between the top of your graph s and the top edge of the Graph sheet Bottom O ffset Specify the distance between the bottom of your graph s and the bottom edge of the Graph sheet Left O ffset Specify the distance between the left most side of your graph s and the left edge of the Graph sheet Right Offset Specify the distance between the right most side of your graph s and the right edge of the Graph sheet H orizontal Spacing Specify the amount of horizontal space to have between multiple graphs side by side on the sheet T his value is used only when you use S PLUS s automatic Arrange Graphs feature to arrange multiple graphs Vertical Spacing Specify the amount of vertical space to have between multiple graphs stacked vertically on the sheet This value is used only when you use S PLUS s automatic arrange graphs feature to arrange multiple graphs Au
250. e name of the class in Classes For more information on object classes see the data class command in the on line help Object classes include the following aov coxph cts data frame dates design factor factanal function gam gim htest its list Im Ime Imsreg loess Itsreg manova matrix multicomp nime nis ordered princomp raov rts smooth spline survfit survreg tree trellis ts vector Include D erived Classes Check this to display not only objects having the specified classes but also objects which inherit from the specified classes See thecl ass command in the on line Language Reference String The object name filter string for the right pane Wildcards such as single character multiple character and ALL are acceptable The ALL wildcard denotes all objects are accepted and is the default Case Sensitive C heck this to impose case sensitivity on the string match 1 a Aa TAT 4 CAPROGRGM FLESVEPLU SA5 FURE ICPROG RA PLES EA DATAE EATOGNAH PLES SPLISI TAT DATASET aca amare FF ipid Garej imar ES cm Tuner Heip Figure 6 5 Specifying an object filter for a Browser Page Below the properties contained in the Page Info tab of the Browser Page dialog are discussed T hese can also be seen in Figure 6 6 Name Specify a name for the O bject Browser page T his will appear on the tab for the O bject Browser page when the O bject Browser window is set up to display text la
251. e of the surface Draw Bottom Draw Top Specify whether to draw the bottom and top of the 3D surface The bottom is the underside of the surface the top is the upper side of the surface Hidden Line Removal Specify whether to have hidden line removal If you choose No overlapping portions of the surface are transparent instead of Opaque Base Style C olor Weight Specify the line style color and weight of the base of the surface T he base is the foundation of the surface If no base is drawn the surface appears to be floating in mid air PLOT PROPERTIES Fills You can specify color draping options for bands or grids on a surface plot You can let S PLUS interpolate shading between two specified colors or you can use a column of colors from a D ata window You can create a surface plot with color draping automatically by selecting data and choosing one of the surface draping buttons on the Plots 3D palette You can specify your own color draping options by creating a surface plot and specifying the following options on the Fills page of the plot property dialog Fill Surface Choose this option to have the surface mesh filled with one or more colors Fill Type Choose a method to specify the colors Choose Solid to specify a single color Choose image colors to use the color scheme specified on the colors page of the Graph sheet property dialog Choose one of the Color Range options to have S PLUS shade the surface between the spe
252. e plotted is greater than the length of the list the line types will cycle through the list Line Width s Enter alist of integers using either the name of an existing S PLUS vector ora comma delimited list If the number of curves to be plotted is greater than the length of the list the line widths will cycle through the list Mark Censoring Times Check this to mark curves at the censoring times Censoring Mark Symbol Enter a list of characters or integers specifying special symbols used to mark the curves Use either the name of an existing S PLUS vector or a comma delimited list such as 1 2 3 or The default is to use at the censored values Size of Marks Enter a positive number used to control the character size of the censor marks Values less than 1 will produce smaller marks while values greater than 1 result in larger marks Log Axis for Y Check this to plot the y axis on the log scale X Axis Style Select the x axis style standard scale or extend the maximum survival time creating an axis with extended labels on the right side of the plot T his is the default e tight create an axis labeled internal to the data values extended create an axis whose numeric labels are more extreme than any data values Scale Factor for X Axis Tick Label Enter a number used to multiply the labels on the x axis For example a value of 365 25 will give labels in years instead of the original days
253. e right side of the error bar drawn N egative to have only the left side drawn Select neither to have only the mean plotted Auto Errors Specify the method to use for computing the error bars Standard Deviation Standard Error Confidence Level or None These results are automatically computed using the row of y values at each x position Specify None to have no error bars drawn only a symbol will be plotted at the mean If you choose Confidence Level you need to specify a numeric value in the Confidence Level field Confidence Level If you specified Confidence Level for Auto Errors specify anumeric value here Auto Means Specify the method to use for computing the mean i e the data point Arithmetic M edian Geometric only if log axes are being used or None If you choose Geometric first the y data are logged then the mean is computed and then the data are unlogged and plotted Choose N one to have only the error bars plotted and not the mean T hese results are automatically computed using the row of y values at each x position Specify the names or column high low close plots data specifications numbers of your X Y Z and W columns TheY column should contain the average or closing data values Z should contain the high data values and W should contain the low data values For high low close plots you can specify standard symbol attributes style color line weight height and frequency and line attributes style colo
254. ect icon choose Properties from the context menu fill out the dialog with the desired default values and click OK Name Enter a name for the Property object To create a Property object a name must be specified Type Select Group or WideGroup to create a group box Select Page to create a tabbed page Select Normal to create any other type of Property object Default Value Enter a default value for the Property object This will be displayed when the dialog opens Parent Property Enter the name of a parent property if any T his is used by certain internal Property objects Dialog Prompt Enter text for thelabel which will appear next to the control in the dialog Dialog Control Choose the type of Property object Examples are Button Check Box List Box and Combo Box Range Enter the range of acceptable values for the function argument associated with this property For instance if the values must be between 1 and 10 enter 1 10 578 DIALOGS Option List Enter a comma separated list The elements of thelist are used for example as the labels of Radio Buttons or as the choices in the drop down box of a String List Box A property may have either a range or an option list but not both Ranges are appropriate for continuous variables O ption lists are appropriate where there is a finite list of allowable values Property List Enter a comma separated list of the Property objects included in the Group box or on the Page
255. ected Axis from the Format menu 2 From the Polar dialog choose the desired page Display Ranges Ticks Labels or Font M ake the desired changes and choose O K Alternatively you can access pages of the property dialog by selecting the axis right clicking and selecting a page from the shortcut menu In the Polar Axes dialog the D isplay page has the following options Axis Display You can specify line style color and weight for the polar axes the line around the outside perimeter of the polar plot Hide C hoose whether to hide or display the axes Circle Spoke Grids You can specify the line style color and weight for both the circular and spoke grids on a polar plot Circular grids are the concentric circles that radiate out from the center of the plot The larger the number of circular lines the denser the grid You can either specify the number of circular grids or the distance between grids Spoke grids are the radial lines crossing through the circular grids T he higher the number of spokes the denser the grid You can specify either the number of spokes or the distance between spokes To avoid crowding at the origin S PLUS draws only half the number of spokes in the innermost circle If you specify the distance between the spokes in degrees S PLUS will round the specified degrees to ensure equal distances between the spokes In the Polar Axes dialog the R anges T icks page has the following options Tick Interval
256. ection Common Plot Properties page 300 for details on fill and color attributes For area charts you can specify standard fill attributes color pattern and background and border attributes color style and weight for each area You can create vertical and horizontal bar charts in S PLUS by selecting data and choosing buttons on the plot palette You can change the direction of the bars on your plot by selecting it and clicking a new button or by changing the Direction option on the Position Fills page of the plot property dialog Standard bar charts For a standard bar chart specify the names or column numbers of your X and Y columns If you specify only one column of data X or Y S PLUS will use an integer sequence 1 n for the missing column of data Grouped and stacked bar charts For a vertical grouped bar chart grouped bar charts specify the names or column numbers of your X and Y columns Each series of bars for example the first bar in each grouping as you move across the Graph sheet must have a y value corresponding to each x value T hus the total number of y values must be the number of x values multiplied by the number of bars in each grouping These y values can be specified with multiple y columns or with one long column PLOT PROPERTIES For a horizontal grouped or stacked bar chart the x and y data specifications are reversed Multiple data columns If your y data are arranged in multiple columns you
257. ed Results Smaller values of Convergence Tolerance will require more iterations while larger values will result in convergence being declared earlier Min Scale for Step Shrinkage Enter the minimum factor by which to shrink the default step size in an attempt to decrease the sum of squares Print Iteration Trace Check here to print a summary of each iteration Use Partial Linear Algorithm Check this to use the Golub Pereyra algorithm for partially linear least squares models Hoalinoer Loa Squavog Regnessice EE Morel Opin re Pimanta Predict Pristina Farei tn Syd Aims P Shoe Curput Ss in F Leng Cuil V Fad aai PD Working Residuais CEJ cancel a ic sf cane Short Output Check here to produce a short summary of the nonlinear fit The summary includes the residual sum of squares the parameter estimates the model formula and the number of observations Long Output Check here to produce a detailed summary of the nonlinear fit This summary includes the mode formula the parameter estimates their standard errors and t statistics the residual sum of squares and degrees of freedom and the correlation matrix of the parameter estimates Save In Enter the name of an S PLus data frame in which fitted values and residuals of the analysis are saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that already exists in data
258. edraw plots automatically after each change is made Redraw N ow Redraw all graph objects Commands W indow Show the Commands window 287 CHAPTER 11 WORKING WITH GRAPH OBJECTS MOVING AND COPYING OBJECTS Using Drag An object can be moved or copied on agraph by dragging the object and Drop To move or copy objects using drag and drop 1 Select the object or objects you wish to move or copy 2 To move the object s click and drag the object s to the new position To copy an object press CTRL while you click and drag To move or copy objects between Graph sheets using drag and drop 1 From the Window menu choose Tile Vertical or Tile H orizontal so you can see both the source and target Graph sheets 2 Select the object or objects you wish to move or copy 3 To move the object s into the target Graph sheet press CTRL while you click and drag the object To copy an object click and drag the object into the target Graph sheet Note If you are dragging multiple dissimilar objects the group of objects can only be dropped on a Graph sheet If you are dragging a collection of objects of the same type you can drop the objects on any target area appropriate for that object type For example if you are dragging several line plots they can be dropped on an axes target on the target Graph sheet H owever if you are dragging several line plots and a symbol they cannot be dropped on an axes target they can only be dropped
259. een selection knobs on the corners of the object 289 CHAPTER 11 WORKING WITH GRAPH OBJECTS EDITING OBJECTS Editing a Single Object 290 Any graphic object can be edited by 1 Choosing the appropriate formatting options from the toolbars or menus 2 Selecting the object and then choosing Selected Object from the Format menu 3 Doubleclicking on the object to display its property dialog 4 Right clicking the object to display its shortcut menu and choosing options To edit using the toolbars 1 Select the object you wish to edit 2 Click on the appropriate button to change the object to the attribute selected for example the Line Color button To edit using the Selected Objects 1 Select the object you wish to edit 2 From the Format menu choose Selected O bject The option name changes to reflect the type of selected object for example Selected Arrow When the dialog appears edit the properties you wish to change Choose OK To edit the object directly 1 Doubleclick on the object you wish to edit The property dialog for the object appears 2 Edit the properties you wish to change Choose OK To edit the object using its shortcut menu 1 Right click on the object you wish to edit The shortcut menu for the object appears 2 Choose the desired option from the shortcut menu EDITING OBJECTS Editing Multiple Objects To edit the contents of a composite object M any commands
260. een these pages O bject D etails Check the detail options which will be listed for objects in the right pane T hisis enabled when the List D etails radio button is selected e Position For S PLUS objects it is the database position in the search list where the object resides For columns of matrices and elenents of lists and data frames it is their position within the parent object 149 CHAPTER 6 USING THE OBJECT BROWSER For toolbar buttons and menu items the Pos field contains their relative position on the toolbar and menu respectively e Data Classis the class of object such as data frame design or Im Inheritance lists the data class of the object and any classes from which it inherits Storage M ode is similar to Data Class but includes for example vector and matrix Dimensions describes the dimension of the object the length of a vector the number of components of a list or data frame or the numbers of rows and columns of a matrix D ate is the date of creation co Ubpri Beer Figure6 4 TheO bject Browse property dialog Right Pane tab When object details are displayed in the right pane objects may be sorted with respect to name size date or other detail by clicking in the corresponding header O bjects displayed in the right pane of the O bject Browser can also be filtered 150 CUSTOMIZNG THE OBJECT BROWSER Customizing Object Browser Pages by string pattern matching Found o
261. eenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the model If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Use Covariance List as Input Check this to use a covariance list as model input instead of a data frame Checking this will enable C ovariance List Covariance List Enter the name of a covariance list to be used as alternative model input T his list must have the form of a list returned by cov wt and cov mve Components must includecenter andcov Acor component will not be used however ann obs Component will be used if present Variables Choose several variables to include in the principal components analysis Formula T he Formula edit box is automatically filled using the variables selected from the Variables drop down box There is no response variable for principal component analysis the formula shows the selected variables additively following a tilde T he formula field may be edited directly Scaling Choose either Covariance M atrix unscaled or Correlation M atrix scaled to have unit variance to define the scaling on which the computation of principal components is based on The default
262. eg Dala Ovary Variance Govariance Components Estimabe s Structure unstructured Parametriation matrid oy Standard Dewation s of Random Eftect s fiviercog ty Ising pi s Time ijcos pi Time 295348 2 095781 _ 1 068676 Figure2 9 TheScript window isglit into program and output panes Report W hen the Commands window is closed and a dialog is launched output is Window before cutting and pasting it into another application Call Im formula ozone temperature data environmental na action na omit Residuals Min 1Q Hedian 39 Max 40 92 17 46 0 8738 10 44 118 1 Somf ficients Value Std Error t value Pr gt t Intercept 147 6461 18 7553 7 8723 0 0000 tamnar atur 7 4391 0 2393 10 1919 Figure2 10 A Report window isan option for holding textual output 44 Imeftalliches Ifsing e Time licps pis Time cluster Mare data Ovary directed to the Report window Text in the Report window can be formatted S PLUS WINDOWS S PLUS Menus The Report window is similar to the Script window T hey both are primarily text windows which can be opened and saved via the File menu and are editable Unlike the script window the report window does not deal with programs or scripts T he report window is a place holder for the text output resulting from any operation in S PLUS Error messages and warnings are sometimes placed in a Report window S PLUS menus change dep
263. egression models via least squares It calls the m function and its print summary plot and predict methods To perform linear regression Choose Statistics gt Regression gt Linear from the main menu The dialog shown below appears Lintar Rogeossion docked Fieni Fin Prei Dem Daia Fiame I H Wet a Subset Rows wih Sea Model Object E Omi fF ath Missing ahes Ga ha fa im Pons Faruka Cae Fords Data Data Frame Select a data frame gt Tip 414 You can typeinto the D ata Frame edit box any expression which evaluates to a data frame Weights Enter the column that specifies weights to be applied to all observations used in the linear regression To weigh all rows equally leave this blank Subset Rows with LINEAR REGRESSION Formula Save Model Object Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to
264. elimiters and spaces and commas can still be used within a character field as long as that field is enclosed within double quotes D ouble quotes cannot be used as standard delimiters If a variable is specified to be numeric and if the value of any cell cannot be interpreted as anumber that cell is filled in with a missing value Incomplete rows are also filled in with missing values If your Excel worksheet contains only numeric data in a rectangular block starting in the first row and column of the worksheet then all you need to specify is the file name and file type If a row contains names specify the number of that row at the Name Row prompt it does not have to be the first row You can select a rectangular subset of your worksheet by specifying starting and ending columns and rows Excel style column names e g A AB can be used to specify the starting and ending columns An application that can support multiple tables or data sets such as Access SigmaPlot SYBASE Oracle Informix Microsoft SQL Server SAS will support exporting multiple tables or data sets to a file S PLUS currently only supports importing the first table from the file unless the file type is O DBC If your Lotus type worksheet contains numeric data only in a rectangular block starting in the first row and column of the worksheet then all you need to specify is the file name and file type If a row contains names specify the number of that row at the N am
265. ells using Cut Copy and Paste 1 2 Select the cells you want to move or copy From the Edit or shortcut menu choose Cut to move cells or Copy to copy cells Or use the keyboard equivalents CTRL C copy CTRL X cut CTRL V paste Click the mouse in the desired location in the D ata window paste area From the Edit or shortcut menu choose Paste If the paste area already contains data the cells will overwrite the existing data cells MOVING AND COPYING DATA To Move or Copy You can use the Copy or Move options on the D ata menu to move or copy Columns and columns and rows within or across Data windows In the dialogs you can Rows specify the source and target locations for the columns or rows you wish to move To move or copy one or more rows or columns 1 Select the rows or columns you want to move or copy 2 From the Data menu choose Move or Copy Choose Row or Column from the submenu 3 The selected rows or columns are automatically filled in for the source specification Specify the target location including the D ata window if different from the source in the M ove or Copy dialog 4 ChooseOK Tneget a Soo fEsrtqueke Tagai Goen Bgus Cod unre a Euske Forcier l Darie Ivg MAE 7 Sowers icuksrvaa Tapco sasac l Orie Figure 4 4 TheM oveand Copy Columns dialogs M ove and Copy Rows are very s milar 101 CHAPTER 4 USING DATA WINDOWS INSERTING AND D
266. elp 19 output grids contour plots 311 output window Preference 601 P parametric survival modeling 521 partial residual plots 418 partial scripts running 549 Paste 100 552 pattern styles 603 Pearson s chi square test 410 permuting observations 353 Picture dialog 165 pie chart labels 325 column of labels 325 justifying 326 offseting 326 precision 325 stacking 326 pie charts 325 colors 325 explode amount 326 font options 326 labels pie chart labels See margin 325 slice options 326 gart angle 325 planenumber 301 planes inserting 201 plot buttons adding plots 183 creating graphs with 177 INDEX plot properties axes numbers 300 borders 302 break at missings 301 break at symbols 301 common 300 connection type 301 cropping 301 dialog 299 fill attributes 302 fills 302 line attributes 301 plane number 301 row range 301 scaling 300 special colors 302 symbol attributes 301 plot property dialog 299 plot type changing with a dialog 296 plots adding data using adialog 187 by axis type 296 formatting multiple 300 replacing data using drag and drop 187 unhiding 186 polar plots data specifications 326 lines 326 symbol 326 text as symbol 326 polynomial regression order curve fitting 315 PowerPoint presentation example 83 PowerPoint presentations 170 precision changing 115 comment plot text 310 contour labels 312 pie chart labels 325 predicted values curve fitting 315 prediction type 463 predictions 4
267. emain menu Main menu Notes Options General settings for options styles and color schemes See chapter on Customizing Your S PLus Session Wi ndow Standard windows controls such as Cascade and Tile Hel p Gives online access to this U ser s Guide and the other manuals common questions and answers and visual demonstrations 46 S PLUS WINDOWS S PLus Dialogs in S PLus user defined dialogs can be linked to S PLus functions The controls in the dialog are mapped to the arguments of the function The function can be displayed in the Object Browser and can be invoked by double clicking on it there It is also possible to access a user defined dialog from a menu item or from a toolbar button Dialogs can contain multiple tabbed pages of options To see the options on a different page of the dialog check the page name When you choose OK or Apply any changes made on any of the tabbed pages are applied to the selected objects Multiple Comparisons Confidence Intervals lel Es Model Options Model Selection Error Specification Model Object Confidence Level 0 95 Name String Match Bounds Jupper and lower gt Variable Selection Error Type iamily wise gt Compare Levels Of o ST Results Compare To Level E Save s flast multicomp m Method Options MV Print Results i mca id Comparisons V Plot Intervals Method best fast Cancel Apply q c
268. emporary default names To edit a script into a Script Window You can type commands directly into a Script window program pane using commands and expressions in the S PLUS Language When you click the mouse in the upper pane the caption or title of the window changes to the name of the script followed by program As you type in this pane the line and column number indicators change to reflect where you are editing The lower pane is used for script output When you run your script all output such as that from the pri nt function calls commands in your script and any warnings or errors normally appear in this output pane this can be changed through the Options menu When you click the mouse in this pane the caption of the Script window changes to the name of the script followed by output You can copy text out of this output pane into the clipboard but you cannot enter text To open an existing script file Ww W Click the O pen button on the standard toolbar Or From the File menu choose O pen Or In the Object Browser select the type of files to be S PLUS Script Files ssc Use the Browser to navigate to the folder of your choice and select the script file for instance my_script ssc THE SCRIPT WINDOW 3 Click on O pen to create a new Script window named my_script ssc Open Look in a samples c i SJmy_script ssc S Script ssc File name Files of type Scri
269. en This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last sample is the most recent sampling result Print Results Check here to print results in the designated output window T he default is not to print output to the screen since output is typically somewhat long Related programming language functions sample set seed Random seed DENSITY CUMULATIVE PROBABILITY OR QUANTILE DENSITY CUMULATIVE PROBABILITY OR QUANTILE This dialog computes density values cumulative probabilities or quantiles from a specified distribution To generate density cumulative probability or quantile Choose Data Distribution Functions from the main menu T he dialog shown below appears Doaaity Comulative Probabibey of Guantila is f Ciena Cala Lhe Daa Freire Greets Sequesce I Deis Frame T vWedable zj Probatebth or Queries Pammer n of the Setected Costnbution l Deriye kirii P Cunnul eva Peokesblirs hiim um fi F Quade Cristi biian union Parameters ofthe Selected Diritti Paral Siem Ae fist pocket T Priv Feweks gt Tip Thedialog can take two input formats Oneisto usa column in an
270. ename To open a toolbar which has been saved in an external file right click on the default Toolbar object and select O pen from the context menu In the O pen dialog navigate to the desired file select it and click OK T he new toolbar is visible in the Object Browser Its name is the name of the external file without the extension STB 575 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE DIALOGS 576 i Jeet Hrnenuert In S PLUS virtually every dialog has an associated object such as BoxPlot XAxisTitle function and so on However all customizable dialogs are associated with functions and they are known as function dialogs T hink of a function dialog as the visual version of some S PLUS function For every function dialog there is one S PLUS function and for every S PLUS function there is a dialog The dialog controls in the dialog correspond to arguments in the function and vice versa H owever all function dialogs are displayed with OK Cancel Apply modeless buttons that do not have any corresponding arguments in the functions When theOK or Apply button is clicked on the function is executed with argument values taken from the current values of dialog controls The characteristics of the controls in the dialog are defined by property objects Filter by Property in the O bject Browser figure 22 6 to see objects of this type CE Ee oe Page 4 1 ot fit T Property apoph Froga Flipa Paopaity Papa Propert
271. ending on the type of document window type Graph Sheet Commands window etc which has the focus the iten may or may not be visible Select the document types for which the item should be visible in the menu system The choice All D ocuments causes the item to be always visible T he choice No Documents ensures that the item will be visible when no document window has the focus for example when no document window is open Action This applies to M enultem objects of type M enultem e NONE No action is performed when the item is selected This is useful when designing a menu system It is not necessary to specify commands to execute when the type is set to NONE BUILTIN One of the actions associated with the default menus or toolbars is performed when the item is selected T hese are listed on the Command page in the Built In Operation drop down box This option allows you to usein a customized dialog any of the intrinsic menu actions such as W indow C ascade e FUNCTION Under this option an S PLUS function either built in or user created is executed when the item is selected OPEN The file specified on the Command page is opened when this item is selected The file will be opened by the application associated to it by the operating system PRINT The file specified on the Command page is printed when this item is selected The file will be printed from the application currently associated to it by
272. ending on which window typeis active W hen you choose one of the main menu options a list of additional options will drop down You can choose from any of the active options in the list M enu options with a gt symbol at the end of the line display submenus when selected Menu items with an ellipsis after the command will display a dialog To access shortcut menus for a selected object simply right click the mouse Shortcut menus contain options specific to the selected object Table 2 2 Themain menu Main menu Notes File File N ew and File O pen create or open documents and windows U se O pen for example to open up the Examples O bject Browser Examples SBF Edit Editing options such as cut copy and paste View Standard options such as whether the status bar and toolbars are visible Insert W hen a Graph sheet is open there are many options for adding graphs annotation and so on see chapters starting with Creating a Graph Can be used to insert folders and browser pages when the O bject Browser is open Format O ptions related to the formatting of objects There are many options available if a Graph sheet is active see chapters starting with C reating a Graph Data D ata editing options See the Statistics section of this guide Statistics Seethe Statistics section of this guide Graph See chapters starting with C reating a Graph 45 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE Table 2 2 Th
273. equence is repeated for each Y value The Y values are also ascending with each value repeated for each X value T his is the format that should be used for creating multipanel graphs Table 8 5 Sample tacked data long form X Y Z 1 2 10 0 60 2 2 10 0 65 3 2 10 0 67 4 2 10 0 69 5 2 10 0 71 1 2 20 0 59 2 2 20 0 62 3 2 20 0 65 4 2 20 0 66 5 2 20 0 67 199 CHAPTER 8 CREATING A GRAPH Table 8 5 Sample tacked data long form 1 2 30 0 55 2 2 30 0 58 3 2 30 0 62 4 2 30 0 66 5 2 30 0 73 Specifying For an irregular surface or contour plot you must specify three columns of Irregular Form equal length for X Y and Z The data can bein any order and the spacing Data between X values and Y values may be random Each X Y Z triplet defines a position in 3D space S PLUS first estimates a set of gridded data and then plots the data as it does a gridded surface or contour plot You will get better results if the data are distributed fairly uniformly in X and Y and do not contain sharp spikes or drops in Z Table 8 6 Irregular data for surface or contour plot X Y Z 1 1 51 0 0 60 2 3 11 9 0 65 3 2 35 8 0 67 3 4 29 0 0 69 4 5 21 6 0 71 5 1 43 2 0 59 5 2 10 3 0 62 200 PROJECTING A 2D PLOT ONTO A 3D PLANE PROJECTING A 2D PLOT ONTO A 3D PLANE Combining M ost of the 2D plot types can be projected onto a 3D plane This can be Multiple 2D useful if you want to combine multiple 2D p
274. er Z and the Z data will automatically be used to draw the error bars no additional calculations will be done Asymmetrical error bars If you wish to draw asymmetrical error bars specify a column containing the values above the mean under Z and a column containing the values below the mean under W For error bar plots you can specify standard symbol attributes style color height line weight and frequency and line attributes style color and weight Additional error bar options are Bar Style Color Height and Line Weight Specify the line style color height and line weight for the vertical or horizontal error bars Cap Width Specify the width of the bar or cap drawn on each end of the error bars in inches cm Specify whether to draw the error bars vertically or horizontally Direction Specify the direction s for the error bars Positive or N egative If your error bars are oriented vertically and you choose both Positive and PLOT PROPERTIES High Low Close Plots Data to Plot Line Symbol Negative the error bar is drawn above and below the mean data point Select Positive to have only the upper portion of the error bar drawn Negative to have only the lower portion drawn Select neither to have only the mean plotted If your error bars are oriented horizontally and you choose both Positive and N egative the error bar is drawn to the left and right of the mean data point Select Positive to have only th
275. er the name of a data frame that already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame T his allows you to keep predicted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used NONLINEAR LEAST SQUARES REGRESSION Options Predictions Check here to store the predictions in the object specified in Save In Confidence Intervals Check here to store lower and upper confidence limits in the object specified in Save In The column names will be N L C L and N U C L where N is 100 times the value specified in C onfidence Level Standard Errors Check here to store the pointwise standard errors for the predictions in the object specified in Save In Confidence Level Enter the confidence level to use when computing confidence intervals T his value should be less than 1 and greater than 0 S PLus language functions related to Nonlinear Least Squares Regression nls print nls predict nls summary nls Other related S PLus language functions ms niminb nlregb 425 CHAPTER 18 FITTING STATISTICAL MODELS LOGISTIC REGRESSION 426 The Logistic Regre
276. ers benefits that traditional language analysis programs simply cant match With S PLUS every data set function or analysis model is treated as an object which makes it easy to 15 CHAPTER 1 WELCOME TO SPLUS Installation 16 examine and visually explore data run functions one step at a time and visually compare models for fit S PLUS gives you immediate feedback because it runs functions one at a time With S PLUS you ve got control over every step of your analysis Visually compare different models for fit reexplore your data for outliers or other factors that might influence a result and document every analysis function Because S PLUS puts you in control you ll have complete confidence in the quality of your results Now standard analysis functions are conveniently available through menus toolbars and dialogs putting powerful S PLUS techniques at your fingertips With point and click ease you can import your data select your statistical functions and display your results As always when your analysis requires a new method or approach you can modify existing methods or develop new ones with the programming language By tapping into the power flexibility and extensibility of S PLUS you can take your analysis to a new level To install the software 1 Insert the CD ROM into yourCD ROM drive 2 If your operating system supports AutoPlay eg Windows 95 or NT 4 0 installation will proceed automatically If not ru
277. ertical Now you can see the data and Graph sheet simultaneously 4 Select the data columns you want to plot Use ctrt click to select discontiguous columns 5 Click the 2D or 3D Plots button on the toolbar A palette of available plot buttons appears 6 sHIFT click the desired plot button on the palette The plot is added to the selected graph using the selected data columns If no graph was selected before sHiFt clicking a new graph will be added to the current Graph sheet When you drag plot buttons from the palette and move them over a 2D graph axes targets appear at the intersections of axes pairs indicating the pair of axes the new plot is to be associated with In the following example there are two sets of axes so there are four possible axes combinations When you drag a plot button over an axes target the target box enlarges If there is only one set of axes dropping a plot button anywhere in the graph will add a plot VIEWING GRAPH SHEETS to that axes pair Drop plot buttons onto axes targets Figure 8 6 Dragging and dropping onto an existing graph To add a plot using drag and drop 1 Click the2D or 3D Plots button to open a plot palette 2 Drag a plot button from the palette onto an axes target at the intersection of the axes and release the mouse 3 A plot icon representing the dropped plot is drawn on the graph 4 Select the data columns you wish to plot Use ctat click to select dis
278. ertical units T his is essential for comparing the importance of fitted terms in additive models S PLus language functions related to Analysis of Variance aov summary aov plot Im coef dummy coef Other related S PLus language functions Im manova raov multicomp 449 CHAPTER 18 FITTING STATISTICAL MODELS RANDOM EFFECTS ANALYSIS OF VARIANCE T his dialog tests if the variations among different factors are the same The effects of factors are assumed to be random To perform random effects analysis of variance Choose Statistics gt Analysis of Variance gt Random Effects from the main menu T he dialog shown below appears Hamidom Liects Anaheis ol Vanaace Madal Gpinra Pemba Pine Deca Dala Fiame tee hls Subst Puras with Seve Model Object Fe Orne Pines mat Mansing ekes SAPTA faris Pomiua Faruka Cae Fimis Model page Data Data Frame Select a data frame gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame Weights Enter the column that specifies weights to be applied to all observations used in the analysis of variance To weigh all rows equally leave this blank Subset Rows with 450 RANDOM EFFECTS ANALYSIS OF VARIANCE Formula Save Model Object Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expres
279. es are from 0 3 to 0 5 For small samples n lt 50 or if there are substantial serial correlations between observations close in x value a prespecified fixed soan smoother should be used Degree O verall degree of locally fitted polynomial One is a locally linear fitting and two is locally quadratic fitting Family T he assumed distribution of the errors T he default is Guassian If Symmetric is selected a robust fitting procedure is used Deg of Freedom The degrees of freedom should be between 1 and the number of input data points 1 The lower the degrees of freedom the smoother the line If Auto is selected cross validation is used Bandwidth The kernel bandwidth smoothing parameter All kernels are scaled so the upper and lower quartiles of the kernel viewed as a probability density area 0 25 when bandwidth is 1 Larger values of bandwidth make smoother estimates Kernel T he choices are Box Triangle a box convolved with itself Parzen function a box convolved with a triangle and Normal the Gaussian density function PLOT PROPERTIES For User Defined Bubble Plots 2D Color Plots 2D Function Name The name of the function to use for smoothing It s first arguments must be x vector of x data y vector of y data z vector of z data can be NULL w vector of w data can be NULL subscripts vector of row indices panelnum panel number if conditioned It must return a list contai
280. es as specified in their corresponding property objects Call Back Function TheS PLUS function backCrosstabs is executed each time a control in the dialog is exited To look at the code behind the function type page backCrosstabs at the prompt in the Commands window A Notepad window opens as is shown in figure 22 14 The highlighted section contains commands which F commanda _ page hackCrosstabs A pm egeSBe SOC Matupad Anoka E n ae data SetEnableFlag data SPrapCroasetRespenge T 2 data lt SetCurrlalue data SPropCrosstFormula data lt SetEnableFlagi date SPropCrosestResponse F data SetCurrlalue data SProapCrosatReaponee Figure 22 14 A look at the code behind the callback function backCrosstabs execute when All Variables is chosen from the Factors drop down box The formula is placed in the Formula edit field TheResponse edit field is disabled The default value of the Response edit field is set to the empty string Write Arg Names T hisis currently empty Display Thisisnot checked so debugging messages will not be shown when the dialog is displayed 585 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE MENUS Creating Menu Items Manahe Objwet 1 All erate Commend hiarr hiomijos Toor Tepe tue Bar Tet Acton C Hie Document Type F Darabi M enus are represented in the O bject Browser as a hierarchy of three type
281. est north south and velocity by a 61 CHAPTER 3 TUTORIAL CTRL click on each column 4 Click the 3D Plot button to open a palette of available 3D plot types see figure 3 3 5 Click on the Scatterplot button shown below in figure 3 9 Flots 30 Figure 3 9 The Scatterplot button on the 3D Plot Palete circled and the 4 Pand Rotation button is shown by the dlips 6 You can interactively rotate your 3D graph Click outside the surface plot but inside the 3D workbox area to select the invisible 3D workbox several green circles should appear D rag horizontally on one of the green circles that appears When you release the mouse the graph is redrawn at the new perspective T he green triangle can be dragged up and down to rotate the graph vertically 7 To see different rotations at the same time use multipanel rotation Click on any point of the plot to select it a single green square should appear Then click on the 4 panel Rotation button on the 62 QUICK TOUR 3D Plot palette see figure3 9 To adjust the starting rotation angles you can interactively rotate the graph in the bottom left panel X Angle 54 Xx Angle 144 X Angle 324 Figure3 10 Four pand rotation four circles and a triangle used for controlling the rotation will appear in the lower left plot Calculating a Tree S PLUS offers a wide variety of statistical techniques including regression Based Model analyses analysi
282. est way to specify these arguments is to work through the Customize dialog available through the context menu for the item in the O bject Browser For details on doing this see the section Using the C ontext M enu page 589 below Show Dialog On Run This is relevant when Action is set to FUNCTION on the Command page Check this to cause the dialog associated to the function to open when the item is selected This can also be specified through the context menu MENUS Modifying Menu Items Using the Property Dialog Using the Context Menu Always Use D efaults T hisis relevant when Action isset to FUNCTION on the Command page Check this to force the use of the default values when the function executes T his can also be specified through the context menu S PLUS makes a distinction between the default argument values for a function as defined in the function s dialog via the Functi onl nfo object and as defined by the function itself Always Use Defaults refers to the dialog defaults Table 22 1 summarizes how Show Dialog On Run and Always Use D efaults work together M enultem objects can be modified using either ther property dialogs or their context menus M enultem objects can be modified through the same property dialogs which are used to create them To modify a Menultem object open the O bject Browser to a page with filtering set to Menultem Right click on the M enultem object s icon in the right pane and
283. f this list send the message unsubscribe to the same address Once enrolled on the list you will begin to receive email To send a message to the S news mailing list send it to S news utstat toronto edu M athSoft Educational Services offers a variety of courses designed to quickly make you efficient and effective at analyzing data with S PLUS T he courses are taught by professional statisticians and leaders in statistical fields Courses feature a hands on approach to learning dividing class time between lecture HELP SUPPORT AND LEARNING RESOURCES S Press Technical Support Books on Data Analysis Using S PLus and online exercises All participants receive the educational materials used in the course including lecture notes supplementary materials and exercise data on diskette S Press is a free quarterly newsletter about S PLUS mailed to primary users of S PLUS S Press features stories by S PLUS users in industry and academia a technical support column and provides new product announcements and other information from M athSoft In North America to contact technical support call 206 283 8802 ext 235 or fax to 206 283 6310 or send e mail to support statsci com In Europe Asia Australia Africa and South America call 44 1276 452299 or fax to 44 1276 451224 or email to shelp mathsoft co uk General Becker R A Chambers J M and Wilks A R 1988 The New S Language Wadsworth
284. factors by default the Helmert contrasts are assigned to unordered factors and polynomial contrasts are assigned to ordered factors to Variable s Select one or more variables to which the selected contrast in Assign Contrast will be assigned Contrasts This field displays the selection and assignment chosen in Assign Contrast and to Variable s 446 FIXED EFFECTS ANALYSIS OF VARIANCE Results page Printed Results Saved Results lined Chocts Anabyais of Sarno Model Cpto Peba Pior Proving Fazy iy Saed Resis F Shot curpul Baw in AROS Tale T Ad lege l Esimajd Goetioerrts V Fiesa F Erimesi k Cogit tor Hoal Fao Short Output Check this to print a short summary of the model fit This includes the model formula the AN OVA table for the model and the residual standard error ANOVA Table Check this to print an AN OVA table for each effect Estimated Coefficients Check this to print the estimated coefficients T here are K 1 such coefficients for each K level factor Estimated K Coef for K level Factor Check this to print K coefficients for each K level factor Fitted Values Check this to save the fitted values Residuals Check this to save the residuals Save In Enter the name of an S PLUS data frame in which fitted values and residuals are to be saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that a
285. fault SBF and is found in your _Prefs directory The O bject Browser toolbar button is linked to the D efault SBF file Packaged with S PLUS are various example data sets To easily view some of these data sets in an Object Browser an examples Object Browser configuration file is provided in your working directory named Examples SBF To open the examples O bject Browser select the File O pen menu item In the O pen File Dialog make sure you are listing Files of type O bject Browser in your working directory Select the file Examples SBF The Find S PLUS O bjects option on the toolbar is a powerful searching tool The dialog takes as input a pattern entered through a drop down list and produces a folder containing references to all objects that match the given pattern Patterns from previous searches are saved and can be recalled using the drop down list T he search goes through all attached S PLUS databases putting the references in a standard Object Browser folder enabling you to search the resulting folder just as you would any other folder This dialog implements thefi nd obj ects function of S PLUS and will only find legitimate S PLUS objects not other items such as files TO FIND S PLUS OBJECTS Shortcut keys Table 6 3 O bject Browser shortcut keys Short cut Left pane Right pane Page Up scroll up scroll up Page D own scroll down scroll down Up arrow move up to next tree item move up to next list iten
286. fferent names T he default is last glm The saved object is of class gI m See the on line help for gl m obj ect for GENERALIZED LINEAR MODELS more information on the saved object Greammahred Linear Models Model Optra Feri fea Fior pmmgeaip Peseta Fd com harati fat Comeagence Tolerance r Frin komin Tiani Options Page Optimization Maximum Iteration Parameters Enter a number specifying the maximum number of iterations If convergence has not been reached after this number of iterations the procedure will stop Convergence Tolerance Enter a number specifying the convergence tolerance Iteration will continue until the relative change in the log likelihood is less than this number Print Iteration Trace 459 CHAPTER 18 FITTING STATISTICAL MODELS 460 Results Page Printed Results Saved Results Check this to print iteration details while the model is being fitted Geammhred Lincar Modets mE A bode Cpto ates Pior Pradis Prriag Fap iy Saved Resis P ihip Ss in l P Leng Ougan l Fiod aai r ANOVA Tohis V Yoring Rendunis I Corelshon kisima Esmaa P Foanman Residuals r Dina Paidia l Response Ge dels Short Output Check this to display a short summary of the modal fit to the designated output window T his includes the model call the degrees of freedom and the residual deviance Long Output Check this to display a detailed summary of the model fit to the designated outp
287. fields determine the range of models examined in the stepwise search Upper Formula Enter the upper formula with the most terms that defines an upper limit for all the models to be tried in the model search Lower Formula Enter the lower formula with the fewest terms that defines a lower limit for all the models to be tried in the model search The default is the NULL model Create Formula Click this to open a formula builder dialog used to construct a formula for either U pper Formula or Lower Formula See the chapter Building Formulas for more information Stepping Direction Select the mode of stepwise search both backward or forward Print a Trace of All Fits Check this to print information for all the fits in the stepwise search Save As Enter the name for the object in which to save the results of the chosen linear model If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he 44 CHAPTER 18 FITTING STATISTICAL MODELS 442 Results page only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names T he default is last step T he saved object will have class I m Seethe on line help for I m obj ect for more information about the saved object Stupwine Linear
288. frame Total Sum Check this to generate the sum of all numeric values in each column of the data frame Summarize Categorical Variables Check this to include summaries of the categorical variables factors in the data frame The corresponding summaries will be the factor levels and a count of how many values in each level are in the factor column Grouping Variables Specify the names of grouping variables to calculate summaries by group Subgroups of data will be formed for each possible combination of the 363 CHAPTER 16 SUMMARIZING AND EXPLORING DATA 364 Results grouping variables M aximum Unique N umeric Values Enter a number Numeric grouping variables having more distinct values than this number will be binned Otherwise one group will be formed for each distinct value of the numeric variable Number of Bins for N umeric Values Enter a number If a numeric grouping variable is to be binned it will be binned into this many bins each of equal width Print Results Check here to display the table of counts in the designated output window Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanu meric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period N ames are
289. frame the text is pasted as column data or e if you arein a script the text replaces currently selected text If you get unexpected results when pasting use the Undo option on the Edit menu You can drag data between S PLUS and other applications that support OLE 2 object linking and embedding version 2 To drag data from another application into an S PLUS graph Start the source application and load the desired file Start S PLUS From the File menu choose N ew Choose GraphSheet and click OK Click on the 2D or 3D Plots button on the standard toolbar A palette of available plot types appears UW LS ore 5 Drag the desired plot button from the palette and drop it inside the Graph sheet D efault axes are drawn and the plot icon is drawn on the graph N ow you need to drop the source application data onto the plot icon to draw the plot 6 Tile the windows vertically to display the source application and S PLUS side by sidein windows 7 Highlight the source data with the mouse Drag and drop the data on the plot icon on the S PLUS graph The icon is replaced with a plot using the source application s data 161 CHAPTER 7 EXCHANGING OBJECTS WITH OTHER APPLICATIONS Creating and Editing Embedded Objects Creating and Editing In Place Creating and Editing in the Server Program 162 You can embed OLE objects directly in an S PLUS Graph sheet without leaving S PLUS T he embedded object can be created as a
290. from its position at the edges of the plot area or from origin if crossed T he default value is 0 meaning that the horizontal and vertical axes meet W hen an offset is specified the plot s will be drawn in the same location but the axis itself will be shifted by the specified amount This feature is especially useful when you are adding multiple axes to the graph Frame Choose whether to have a frame or mirrored axis on the opposite edge of the plotting area Choose from N one No ticks With ticks or With labels amp ticks Panel Frames Choose whether to have a frame or mirrored axis on the opposite edge of the panel when you have a multipanel page Choose from Auto No Frame Alternate Sides with Labels O uter Sides with Ticks O uter Sides with Labels or Frame O uter Sides Vary Axis Range Allow the range of the axis to differ between panels of a Trellis plot In the Axis dialog the R ange page has the following options Axis Range Specify the minimum and maximum values of each axis in axes units or the units of the data in the plot Axes units are interpreted just like x and y data points on the graph T he default Auto produces a range outside the data values and typically produces a small margin at both ends of the axis DataM in and DataM ax specify that the minimum or maximum data value be used for the axis For example if you specify DataMin for the Axis Minimum and D ataM ax for the Axis M aximum your axis range
291. g 2D Axis Labels page 244 for information on other label type options Precision Specify the number of displayed digits after the decimal Column If Column was chosen for the Label style specify the name of the column containing the slice labels D ata Set Specify the name of the data set containing the column of labels 325 CHAPTER 12 EDITING PLOT PROPERTIES Font By Slice Polar Plots Data to Plot Line Symbol 326 Stacked Choose this option to have the two labels stacked vertically Label 1 on top If this option is not selected the two labels will be placed side by side Label 1 first Label 2 is automatically enclosed in parentheses if the label is a number Offset Specify the amount of offset between the labels and the center of the pie Negative numbers will move the labels toward the center of the pie positive numbers will move them farther away from the center of the pie A negative offset produces the best results when pie slices are large and fairly evenly spaced Justification Specify the justification Auto Right Left Corner and C enter for stacked labels For all labels you may specify font color size and styles Specify the border and fill attributes for each slice You can make color or style changes to all slices at once or specify properties for individual slices See the section Common Plot Properties page 300 for details on changing fill and border attributes O ther options includ
292. g Specify a scaling type to be used after the axis break For multiple breaks specify a list of scaling types delineated by commas Auto can be used to specify the same scaling type on all axis segments Style Choose a style for the break lines from None Perpendicular 45 D egrees or Wavy Break Length Specify a length for the break lines Gap Width Specify a width for the gap between the break lines Color Choose from a list of colors for the break lines Line Weight Choose from a list of line thicknesses for the break lines To select 2D axis label w Click anywhere on the labels to select them You will see a triangular selection knob To move 2D axis labels 1 Select the axis labels 2 Drag the labels inside or outside the axes by dragging the triangular selection knob or 3 Doubleclick the labels or select the labels and choose Selected Axis from the Format menu Choose the Position page and specify values in the H orizontal and Vertical O ffset fields Choose OK To format 2D axis labels 1 Doubleclick the labels or select the labels and choose Selected O bject from the Format menu FORMATTING 2D AXES 2 From the Axis Labels dialog choose the desired page Type Position or Font M ake the desired changes and choose OK Major Label 1 You can specify two rows of major labels in S PLUS using the Label 1 and Label 2 and Label 2 pages of the Axis Label dialog Use the M inor Labels page to specify Min
293. g a Printer Selecting a Plotter Printing Sheets and Scripts Printing Sheets and Scripts 337 Sending Electronic M ail 338 Exporting Graph Sheets to D ifferent File Formats 339 Before you can print your graphs in S PLUS you must specify the printer or plotter your system will be using your printer selection can be changed at any time First use the Windows C ontrol Panel to install a printer or number of printers Once a printer is installed you can change printers easily with Print Setup in S PLUS For more information on installing printers see your Windows documentation S PLUS uses standard Windows Print Setup and Print dialogs To select a printer 1 From the File menu choose Print Setup 2 Select a printer from the list of printers installed on your system 3 Choose the OK button to save your printer selection and exit Print Setup In order to print to a plotter the plotter driver must first be installed using the Windows Control Panel using the same method as installing a printer driver To change specific plotter specifications choose Print Setup and click on the O ptions button You can print S PLUS Graph sheets D ata windows Report files and scripts using the Print button or the Print dialog 337 CHAPTER 13 EXPORTING GRAPHS PRINTING AND SENDING MAIL Using the Print Button w Click on the Print button on the Standard toolbar S PLUS will immediately print the currently selected window using the
294. g a Data To edit a data set in a Data window Set 1 Open the Browser 2 Filter for the data type you are interested in Refer to the chapter on the O bject Browser for more details 3 Double click on the data set you wish to edit to view it in a Data window The Current The data set that last had the focus in the Object Browser or in a Data window is called the current data set It is also the default data set which is Data Window used when no data set is explicitly referenced in an operation To change the current data set click on the data set you wish to makecurrent or select it from the list in the Window menu 89 CHAPTER 4 USING DATA WINDOWS SELECTING CELLS COLUMNS AND ROWS Selecting Cells Extending the Cell Selection Selecting Columns 90 In S PLUS you can select a single cell or a group of cells in a D ata window You can select blocks of cells rows or columns by clicking and dragging the mouse To select a single cell w Click in the single cell to make it active To select a block of cells 1 Position the mouse over the first cell you want to select 2 Hold down the left mouse button and drag the cursor increasing or decreasing the size of the block The blocked area is highlighted W hen the desired area is highlighted release the left mouse button to select the block You can extend the cell selection of a block by holding down the sHirT key while pressing the direction keys For exa
295. g leaves the cells empty removing a column will ddeteit and shrink the ze of the data set When you clear a column the data are deleted and replaced with NAs The column s position name and formatting information remain in the Data window You can also use the Clear option on the Data menu The Clear Column dialog lets you specify complex column lists for example 1 3 5 8 and 104 INSERTING AND DELETING CELLS ROWS AND COLUMNS different D ata windows Note W hen you use the Clear command on the Edit menu the data are not placed in the clipboard Use the Cut command if you want to erase data and place it in the clipboard To remove rows 1 Select the rows you want to delete 2 From the Data menu choose Remove then Row Or from the Remove Rows dialog choose OK Clear Rows Pe Count EEEE germo gu i f Figure 4 8 TheClear Rows and Remove Rows dialogs 105 CHAPTER 4 USING DATA WINDOWS SORTING DATA 106 Son Humit Iig LS Boi Bee Cod uve fiistence Jot Goluma fasience l deceading Figure 4 9 The Sort Columns dialog S PLUS s Sort option lets you sort one or more data columns numerically or alphabetically Your columns can be sorted according to the data in other specified columns T he Sort option sorts data in ascending or descending order You can specify the columns to sort where to put the results in the same or a different D ata window
296. g shown below appears el i Teri mm Dees Hypahaues M Deta Has a Giro ui Wariniie NulHypah Mosa i Ciia Frome Aieri Hyepoth ticked w eure Contd esos lintereal ye Veriabhe g Cowmicharrta Lel fo m Taat Fiale Typa di Tast m Paradi Eman Ai Pisstttiest a Terena E Print fResets i kanimo Equal Vaiano Data Has a Grouping Variable Check this if one column in the data frame is a grouping indicator that categorizes cases into two groups In this case select the response from Response Variable and the indicator from Grouping Variable Data Frame Select a data frame x Variable Select a column as the firs sample when the data frame does not have a grouping indicator y Variable Select another column as the second sample when the data frame does not TWO SAMPLE T TEST have a grouping indicator gt Tip You can have a data frame containing a respons variable and a grouping variable or a data frame with two samplesin two columns Test Hypotheses Confidence Interval Results Type of t Test Choose Paired t for a paired t test choose Two sample t for an unpaired t test Equal variances This is enabled only when Two sample t is chosen Check here if the two samples are assumed to come from populations with equal variances In Null Hypoth Mean Enter the difference between the assumed population means of Variable x and Variable y Alternative Hypoth Selec
297. ge contour plots 311 data sheets formatting 111 saving defaults 111 smart cursor 599 data sheets exporting 136 data specifications editing for multiple plots 189 data types changing 114 numeric formats 114 vector plots 327 Data window defaults 111 entering data 96 D ata windows 86 formatting 111 D ataFrame toolbar 87 dBase files 128 DDE requests 598 decimal places displaying 115 default O bject Browser 55 146 defaults for columns 115 saving data sheet 111 degrees of freedom 378 608 deleting blocks 105 columns 104 plots 186 rows 105 deleting objects 156 delimiters exporting 137 density values 355 description columns 113 dialog execution common controls 341 dialog fidds 342 automatically filled 342 dialog rollback selective undos 108 display format changing 114 comment plot text 310 contour labels 312 Display H istory dialog 559 displaying decimal places 115 draft mode viewing 601 drag and drop adding plots with 184 creating agraph 179 drop lines 3D regression 330 editable graphical objects 602 editable graphics 343 editing data 97 embedding objects 161 end column when importing 121 end row when importing 122 ending entry history log 559 environmental data frame 78 EPS exporting to 340 error bar calculations bar charts 306 INDEX error bar plots 316 asymmetrical bars 316 auto errors 317 auto means 317 bar direction 316 cap width 316 confidence level 317 data spec
298. gend dialog from the shortcut menu In the Legend Box dialog the Position page has the following options Number of Items Specify the number of legend items you want to have If you leave this field at Auto the number of legend items is determined by the number of plots on the graph and will change as you add and delete plots Font Size Specify the font size to be used for all legend items x Position and y Position T he legend position is determined by specifying the distance from the bottom left hand corner of the graph area to the upper left hand corner of the legend box ADDING TITLES AND LEGENDS Fill Border Page Box Specs Page Formatting the Legend Items Hide Choose whether to display the legend Use Axes Units Choose whether the position specification should be interpreted in document units or axes units X Axis and Y Axis f you choose Use Axes Units these options become available If you have more than one x or y axis on the graph these options let you specify which x axis and which y axis you want used for determining the axes units You can specify standard fill attributes fill color fill pattern and pattern color and standard line attributes style color and weight for the box containing the legend items In the Legend dialog the Box Specs page has the following options Sample Length Specify the length of the sample line symbol or pattern in document units Vertical Spacing Specify th
299. ges allows you to specify your own ranges in a D ata window column Lower U pper Range Column Specify columns of lower and upper bounds Skip Column Specify panels to be skipped or drawn as a column of Os and 1s in the Data window for example to specify a column of four panels that skips the third one enter c 1 1 0 1 into aD ata window column Vertical H orizontal Margin Column Specify column of numbers to use as variable vertical and horizontal margins between panels Panel O rder Column Specify a column of numbers to use as the order of the panels For example a column containing c 2 4 1 3 would draw the first panel in the second position the second panel in the fourth position and so on 213 CHAPTER 9 WORKING WITH TRELLIS GRAPHICS Formatting You can have a strip drawn at the top of each panel representing each Panel Strips conditioning variable If the panel is discrete the strip indicates the value of 214 the conditioning variable If the panel is continuous it gives the name of the variable In both cases a shaded band can be used to indicate the relative value or range of the conditioning variable If you do not want shading in your panel strips make the fill color and highlight color the same D efault values for the colors used in the strip labels can be modified on the Fill Color page of the O ptions Graph Styles dialogs 9 99 09 09 99 09 299 99 09 09 99 09 99 KPRROReoQeoh Ra RarFaohr a
300. ght click and select Show Dialog from the pop up menu The dialog for the line plot or histogram will appear Fill it in click OK and your call using the gui Create function is inserted automatically for you in the Script window The class name to use for particular plots and operations is available in the documentation for each plot or operation It is also possible to generate the plot from the user interface and look at the History log for the class name that was used The list of available class names can be obtained interactively by using guiGetClassNames in the Commands window or from a Script window The list of possible arguments for gui Create Or gui Modify forthe particular class name is available by calling gui Get Argument Names THE SCRIPT WINDOW Expand Inplace Font Help your classname To get the dialog for an S PLUS object select the object by double clicking on its name right click and select Show Dialog from the pop up menu T he dialog for the object will appear Fill itin click OK and your changes for this object will take effect immediately If you need to get the body of a function in your Script window select the function name double click on the function name right click select Expand Inplace and the function body is inserted automatically for you in the Script window If the font of the Script window is too small or too large right click in the Script window select Font and the common fo
301. gion you want to format in the lower section of the dialog The current specifications for the selected area are filled into the fields in the upper section of the dialog 3 Make changes to one or more of the fields As you change the values in the fields the specifications are updated automatically in the lower section of the dialog 4 Click Apply or OK to apply the changes to the plot To format more than one item at atime cTat click on the specifications for each item you want to format The fiddsin the upper section of the dialog go blank if the current values in the fields are not the same for each item As you make changes to the fields all selected items will update automatically Use the Select All button to select all items for editing and set an attribute for all items at once Specify the names or column numbers of an x and y column of data If a single y column is specified an x y curve is drawn and the area beneath the curve is filled If multiple y columns or a stacked y column are specified a curve is drawn for each set of values and the area beneath each curve is filled Specifying Multiple Y Columns If your y data are arranged in multiple columns you can specify the label column for X and multiple Y columns for the Y data specifications for example Samplel Sample5 Each column must have the same length See section Preparing D ata for Graphing page 194 for an example of multiple column data Specifying Stac
302. green knobs and a vertical rotation knob a green triangle appear on the workbox 2 Dragon any of the knobs to rotate the workbox 3 When you release the rotation knob the graph is redrawn at the new workbox orientation Figure10 5 Rotating a 3D plot The circular knobs rotate the workbox horizontally around the z axis The triangular knob rotates the workbox vertically around the x axis In a multipanel 3D graph you can interactively rotate the workbox in the first panel All panels will be updated To rotate a 3D graph using the dialog 1 Doubleclick outside the plot area but inside the graph area to display the Graph dialog 2 Choose the 3D Workbox page 254 ROTATING A 3D GRAPH 3D Workbox Page 3 Specify the Angle to Z axis Angle to X axis and Distance and choose OK T he observer s viewpoint is specified in spherical units The viewpoint must be outside the workbox and should be above the XY plane and not directly facing any of the sides of the workbox If the specified viewpoint is below the XY plane the labels are displayed backward In the Graph dialog the 3D Workbox page has the following options Note that the 3D Workbox pageis only available when you are editing a3D graph Angle to Z axis Specify the angle between 0 and 360 degrees measured from an imaginary line that runs parallel to the z axis through the center of the workbox Angle to X axis Specify the angle between 0 and
303. h Jr Model Dhed i yi gis E Om ices wet Manning alen Ag N Saaie hed abe Sair UppaiFamua Lia Formula PAL Crna Femila gt Tip You can typeinto the D ata Frame edit box any expresion which evaluates to a data frame 440 Enter the column that specifies weights to be applied to all observations used in each linear regression To weigh all rows equally leave this blank Subset Rows with Enter an S PLUS expression which identifies the rows to use in each analysis STEPWISE LINEAR REGRESSION Models Scope Stepping Options Save Model Object To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Also see the on line help for the function st ep Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the model If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal T hese
304. h multiple Graph sheets Graphs are object oriented meaning elements can be selected and edited to create publication quality output T here are various ways to create a Graph sheet 1 By selecting data displayed in the Data window and clicking in a plot palette 2 By selecting columns of data from the O bject Browser and clicking in aplot palette 3 By selecting graphing options from the menus 4 By calling functions from the Commands window 41 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE GS1 5 By running scripts of S PLUS commands in the Script window LOL x Paget Figure2 7 A Graph het window displaying a Trellis graph Commands Window 42 To edit a Graph sheet double click on the plot object you wish to modify Specify the edits in the Property dialog which now appears To customize your graph you can select objects directly on the Graph sheet or select the objects from the O bject Browser The appropriate dialog appears allowing you to edit text color line weight and many more similar plotting features The Commands window behaves similarly to that in S PLUS 3 3 for Windows TheC ommands window allows you to access the powerful S PLUS programming language you can modify existing functions or create new ones tailored to your specific analysis needs The new Bootstrap routines along with some other advanced modeling procedures are not available through the Sta
305. h 50 years and many analyses an important happening in the data had gone undetected T he basic analysis follows Creating the Trellis graph 1 Select the barley data frame in the left pane of the O bject Browser In the right pane click on yield then CTRL click on yield variety year and site The order of the variables selected determines how they will be used in the color plot The first variable yield will be used as x the second variety as y and the third year will be used TRELLIS EXAMPLES to determine the color of the symbols T he last site is used as the conditioning variable 2 Open the 2D plot palette Turn conditioning mode on by pressing the conditioning mode button in on the Standard toolbar and make sure that the number of conditioning variables on the Standard toolbar is set to one N otice that there are now yellow strip labels at the top of each of the plot icons 3 Click on the Color Plot icon see the palette in section M ethods for Creating a Graph page 177 A Trellis graph will appear showing barley yield for each variety conditioned on the site The yields for 1931 and 1932 appear in different colors 4 Turn conditioning mode off maximize the new Graph sheet and close the 2D plot palette Formatting the panels To make it easier to compare yields across sites we will make three changes to the layout of the panels stack them in one column reorder them according to the median of the yiel
306. h to collapse the predictions The program will sum the predictions for each level of this variable Cov Proportional Harerde i y Hl Model Optra Fiais Predici Eurial Cursa Pint Curae phions VD Surer Cuns Lira Colorie fi Cun Typs Lire Typeset fi Hew Deis Line isti js fi Daia Frere Agta Options M Only Ong hiiia r Lag Aceon Confidence barsi Mu Sado risida P Ehr Cos idee bine Seaia Factor kor stems Tick Labat Cot dood inpro Lra fi pas Sasis Fasiri Aadi Tick Labada Coridece neral Type fi eg hada Laine G Irt a E coe om a Survival Curves Use the Survival Curves page to plot the predicted survivor function for a Cox proportional hazards model Plots Survival Curve Check this to create plots of survival curves for the current cox regression object Curve Type Select the type of survival estimate aalen or kaplan meier The Aalen estimate of survival which is equivalent to the Tsiatis estimate is used by default New Data Data Frame Enter the name of a data frame with the same variable names as those that 518 COX PROPORTIONAL HAZARDS Confidence Intervals Curve Options Axis Options appear in the formula on the model page The curves produced will be representative of a cohort whose covariates correspond to the values in this data frame If left blank the survival curves will be based on the mean of the covariates used in the cox regression fit Only One Individual Chec
307. h to experiment with different layouts and to rearrange the graph order Note T he default layout and the margins for the Graph sheet and the horizontal and vertical spacing between the graphs can be changed in the Format Sheet dialog You can specify whether to automatically resize fonts and symbols when graphs are resized in the O ptions Graphs O ptions dialog To exchange two existing graphs 1 2 Select the two graphs you wish to exchange using sH FT click to select multiple objects From the Format menu choose Exchange Graphs To arrange new graphs automatically 1 2 3 5 O pen anew Graph sheet From the Format menu choose Sheet In the Auto Arrange field specify the desired layout style N one One Across Two Across T hree Across Four Across or O verlaid This determines the number of graphs to place horizontally before wrapping graphs to the next row Choose OK Each time you add a graph to the page it will be automatically arranged according the selected layout style For example if you choose One Across the first graph you add will be full size If you add a scond graph the first graph will be automatically resized and positioned to accommodate the second graph For more information on the Auto Arrange field see section Formatting a Graph Sheet page 228 193 CHAPTER 8 CREATING A GRAPH PREPARING DATA FOR GRAPHING Some plot types require data to bein a particula
308. hahaha rye n o 2 5 5 gx 0 e fe O 72 2 x o o 2 o a O e e I E aN Ae Rea Per Capita Income Figure9 4 Pand sripson a multipand plot To remove panel strips w Uncheck the Draw Strip check box on the M ultipanel page of the Graph Properties dialog To edit panel strips 1 Doubleclick on the panel strip 2 Change the attributes on the pages of the Panel Strips dialog and CREATING TRELLIS GRAPHS Fills Page Labels Page Fonts Page click OK On the Fills page you can specify the fill and border properties for the panel strips Highlight Color Specify a color to be used to represent the range or value of the conditioning variable for that panel Label Type Choose from N one Values and Column If Column is selected you must specify a Column name and D ata set in the following fields Label Position Choose from Centered On Highlight and Label All Values Format Type C hoose from D ecimal Scientific M ixed and Auto See section Formatting 2D Axis Labels page 244 for information on options Precision Specify the number of decimal places to display after the decimal On the Font page you can specify the font size color and styles for the panel strips 215 CHAPTER 9 WORKING WITH TRELLIS GRAPHICS TRELLIS EXAMPLES Example 1 Pollutants in Automobile Exhaust 216 D ata were collected to analyze oxides of nitrogen in au
309. hanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names Print Results Check this to print the results in the designated output window Related S PLus language functions chisq test print htest menuChisquare 411 CHAPTER 17 COMPARING SAMPLES 412 FITTING STATISTICAL MODELS Introduction Regression Linear Regression 414 N onlinear Least Squares Regression 421 Logistic Regression 426 Log linear Regression 427 Robust LTS Regression 428 Local Regression 433 Stepwise Linear Regression 440 Analysis of Variance Fixed Effects Analysis of Variance 444 Random Effects Analysis of Variance 450 Advanced Models Generalized Linear M odels 456 Generalized Additive M odels 465 Tree M odels 473 Other Methods M ultiple Comparisons 480 Compare M odels 485 S PLUS has many functions to fit statistical models This chapter describes dialogs to run linear models non linear models generalized linear models and many other robust and non parametric approaches for modeling T here are also dialogs to compare models and to perform mulitple comparisons Refer to chapter 14 Using the Statistics Menus and Dialogs for an explanation of common features and also to chapter 20 Building Formulas 413 CHAPTER 18 FITTING STATISTICAL MODELS LINEAR REGRESSION This dialog fits linear r
310. he Fill Color page has ten fill color fields Enter the fill colorsin the order in which you want them to cycle There are an additional ten colors used for strip labels in multipanel plots To specify color schemes select Color Schemes from the O ptions menu T he Color Schemes dialog appears T here are eight available color schemes for User Colors and Image Colors These are used when defining graph styles in the Options Graph Styles menu Set the background color for each color scheme To edit the User Colorsfor a color scheme click on the Edit Colors button and use the color dialog to 603 CHAPTER 23 CUSTOMIZNG YOUR S PLUS SESSION Image Colors Page Redrawing Plots Automatically 604 modify the colors The user color scheme specified in the O ptions Graph Styles of the default Graph Style will be used to set the user colors in any newly created graph sheet T hese colors will appear in the color lists for all of the graphical objects within the graph sheet as User1 U ser2 etc of Colors mage colors are a series of fill colors that can be used for draped surfaces flooded contours and levels plots T he specification of image colors consists of up to sixteen core colors and a list defining the number of shades or color gradations between each core color Number of image colors indicates how many core colors are used in the image colors definition of Shades is defined by a list of numbers separated by commas ind
311. he Graph sheet in which you wish to add agraph M ake sure nothing on the Graph sheet is selected If a graph is selected this procedure will add a plot to the selected graph instead of adding a new graph O pen a Data window containing the data to plot or view it in the O bject Browser From the Window menu choose Tile Vertical Now you can see the data and Graph sheet simultaneously In the D ata window select the data columns you want to plot Use CTRL Click to select discontiguous columns Click the 2D or 3D Plots button on the toolbar A palette of available plot buttons appears SHIFT Click the desired plot button on the palette PLACING MULTIPLE GRAPHS ON A GRAPH SHEET Combining Graphs from Multiple Graph Sheets The graph is added to the current Graph sheet and a plot is placed on the graph using the selected data columns To add a graph using the Insert Graph option 1 From the Insert menu select Graph 2 Specify the desired graph and plot from the Graph Type and Plot Type lists 3 Specify the name of the Graph sheet to which you wish to add anew graph from the Graph Sheet list Choose OK to closethedialog T he graph appears To add a graph using drag and drop 1 Open a Data window containing the data to plot or view it in the O bject Browser 2 Open the Graph sheet in which you wish to add a graph 3 From the Window menu choose Tile Vertical Now you can see the data and Graph sheet simult
312. he H elp button is pushed Write Argument Names Check this to have argument names written when the function call is made Display Check thisto cause information about theFuncti onl nfo object to be written in a message window or in the output pane of a script window when the dialog is launched by a script This debugging tool is turned off after O K or Apply is clicked in the dialog Internal Resources H ere are several internal property objects that can be used in dialogs either alone or by means of Copy From TXPROP_DataFrames This Property object displays a drop down box listing all data frames filtered to be displayed in any browser TXPROP_DataFrameCol umns T his Property object displays a drop down box listing all columns in the data frame selected in TXPROP_DataFrames If no selection in TXPROP_DataFrames has been made default values are supplied TXPROP_DataFrameCol umns ND This Property object displays a drop down box of all columns in the data frame selected in TXPROP_DataFrames If no selection in TXPROP_DataFrames hasbeen made default values are not supplied TXPROP_SplusFormul a T his Property object causes an S PLUS formula to be written into an edit field when columns in a data sheet view are selected T he response variable is the first column selected and the predictor variables are the other columns TXPROP_WideSplusFor mula This Property object differs from TXPROP_ Spl usFor mul a only in that the form
313. he Z data will automatically be used to draw the error bars no additional calculations will be done Draw Error Bars Specify whether to display error bars on each bar in the chart PLOT PROPERTIES By Bar Box Plots Data to Plot Auto Errors Bars Specify the method to use for computing the error bars Standard deviation Standard error C onfidence Level or None T hese results are automatically computed using the row of y values at each x position Specify None to have no error bars drawn only a bar will be plotted at the mean If you choose C onfidence Level you need to specify a numeric value in the Confidence Level field Confidence Level If you chose Confidence Level for Auto Errors specify a numeric value here Auto Means Specify the method to use for computing the mean i e the data point Arithmetic M edian Geometric only if log axes are being used or None If you choose Geometric first the y data are logged then the mean is computed and then the data are unlogged and plotted These results are automatically computed using the row of y values at each x position Cap Width Specify the width of the bar or cap drawn on each end of the error bars in inches cm Specify the border and fill attributes for each bar You can make color or style changes to all bars at once or specify properties for individual bars See the section Common Plot Properties page 300 for details on changing fill and border attributes
314. he amount of offset is a fraction of a single bar width The default offset value of 1 places the bars side by side and touching one another an offset value of 0 places the bars on top of each other Bar Base Specify whether to position the base of each bar at zero or at the minimum value on the y axis For bar charts you can choose a solid color for each bar fill the bar with a series of shades determined by a range of colors or specify new colors for the filled bars using the standard color dialog See the section Common Plot Properties page 300 for details on fill and color attributes Automatic error bar calculation For automatic vertical error bar calculations you need asingleX column and a group of Y columns The X column determines the positions of the bars along the x axis The Y data are used to compute the mean and size of the error bar The height of the bar will be the mean of the corresponding y values A series of Y columns can be specified either in alist for example Y1 Y2 Y3 or in a sequence for example Y 1 Y3 The mean and error value for each error bar are automatically calculated using each row of y values corresponding to each x A bar is plotted at the mean and an error bar is drawn above the bar Thex and y data specifications are reversed for horizontal barcharts No error bar calculation If you already have a column containing the standard error for your data you can specify the column under Z and t
315. he checkbox empty to permit the plot to extend beyond the plot area If cropping is on any data lines or symbols extending beyond the plot area are not plotted Check the Hide option to temporarily prevent a plot from displaying on a graph W hen you hide a plot all of the plot specifications are retained A plot button remains on the graph representing the hidden plot You can unhidea hidden plot by double clicking on its plot icon and deselecting the Hide checkbox Subset Rows with You can also specify a subset of rows for graphing in the Subset Rows with Lines attributes Symbols attributes field Enter an S PLUS expression which identifies the rows to use in the analysis The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of rows to use for example 1 20 or Age gt 13 Line Style Specify the line style for example dotted dashed etc to be used or None to have no line drawn Line Color Specify the color for a line Choose from 16 preset colors or your own custom colors Line Weight Specify the line weight in points Connect Type Choose from a list of line connection types Break Line at Missings Choose whether to break the plot line when it encounters a missing value If this option is not selected the plot line connects all data points ignoring missing values Break Line at Symbols Choose whether to break the plo
316. he column smaller or right to make the column larger You can also specify the exact column width according to the number of characters in the default font and point size in the Column dialog 112 FORMATTING DATA WINDOWS Changing a Column or Row Name or Description To change column width with the toolbar 1 Select the data columns you wish to resize 2 Click the Increase Column Width or Decrease Column Width buttons Each click increases or decreases the column width by one character To adjust column width to the widest cell 1 Select the data columns you wish to resize 2 Click the Column W idth to Fit button Column names are used to refer to data for graphing or data manipulations It is good practice to start column names with a letter and the names may contain any combination of letters numbers or periods T he same name cannot be used twice within a data set If you do not specify a column name the column can still be referred to by its number S PLUS function names and other reserved words cannot be used as column names Column names are automatically used as the default axes titles and legend text in graphs If you specify a column description it is used for axes titles and legend text instead If no column names or descriptions are specified the column number is used for the legend text along with the Data window name for example DF1 1 A column description can contain up to 75 char
317. he data This will affect the number of degrees of freedom associated to the test statistic Hypothesized Dist Select the hypothesized distribution for the variable W hen a hypothesized distribution is chosen the parameters associated with the distribution are enabled For example if the Cauchy distribution is chosen the Location and Scale are enabled Enter the location and scale parameters of the C auchy distribution Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last lm is the most recent linear regression model fit Print Results Check this to print the results of the chi square goodness of fit test in the designated output window Related S PLus language functions chisq gof ks gof qqplot 387 CHAPTER 17 COMPARING SAMPLES TWO SAMPLE T TEST 388 Data T his dialog performs a two sample t test on data held in a data frame To perform two sample t test Choose Statistics gt Compare Samples gt Two Samples t test from the main menu T he dialo
318. he data before drawing Drop Line Attributes Specify whether to have lines drawn from each data point to a specified plane and specify standard line attributes color style and weight Specify N one for the Line Style to have no drop lines drawn Drop lines help viewers distinguish which points lie above the specified plane and which lie below Base Line Level Specify the Z level of the drop line plane For example if your Z data ranged from 10 to 10 you could specify a baseline level of zero to see how many data points lie above and below the zero plane The dropped line for each data point is drawn to a length that reaches the specified base line level D rop lines are automatically drawn to the specified plane For 3D line and scatter plots you can specify standard symbol attributes style color height line weight and frequency PLOT PROPERTIES Vary Symbols Regression Use Text as Symbol C hoose whether to use a user specified string or column of text in any font as plotting symbols Text to Use C hoose from Specified Text x y z w or O ther Column Choose Specified Text to type text into the Symbol Text box and use it for every symbol on the graph Choose one of the column options to use a column of text as symbols Symbol Text Specify the text to be used as a symbol Column Specify a column of text to use D ata Set Specify the data set containing the column of text Font Choose a font type for the text
319. he list If you are going to use this option ignore references in the examples to using the Find button QUICK TOUR QUICK TOUR oi beat PH Search 1 5 a 5 5 G amp G 6 5 5 5 D Ti T Figure3 2 TheExample O bject Browse with data frame highlighted in the left pane First let s open the data in a convenient D ata window 1 Click on the Find button in the O bject Browser toolbar Type in fuel frame for the Pattern and click OK Go to the left pane of the O bject Browser and click on the in front of Found O bjects to expand the list Double click on fuel frame to open a Data window M aximize the D ata window Select all the columns of fuel frame from Weight to Type by dragging the mouse across the column headers Weight Disp M ileage Fuel Type automobile weight engine displacement 6 liter 8 liter etc mileage in miles per gallon 100 mileage category ofvehicle Large M edium Small Compact Sporty Vans Click the 2D Plot button on the Standard toolbar to open a palette 57 CHAPTER 3 TUTORIAL of available 2D plot types Figure 3 3 The Standard toolbar showing the postion of the Object Browser button circled and the2D and 3D Plot palette buttons marked by the dli pse 6 Click the Scatter Plot M atrix button on the palette Plote20 Figure 3 4 The2D plot palate owing the Scatter Plot M atrix button circled and the Linear F
320. he tables in the data source can be determined and the list of table names is initialized The table name defaults to the first table in the data source W hen a table name is selected a default SQ L Query and a default data frame name to import the data into are generated SQL Query Structured Query Language SQL is a powerful flexible and standardized language for extracting data from databases Any legal SQL statement can be entered in this box The default query is generated when you select a table name It selects for import all the data from the table Within the Import To group are the following options Data Frame Specify the data frame to contain the imported data You can specify the name of an existing data frame or a new data frame If the specified data frame does not already exist S PLUS automatically creates it If the specified data frame does exist you will warned that you could overwrite data 133 CHAPTER 5 IMPORTING AND EXPORTING DATA 134 Target Start Col Specify the starting column for the imported data T he start column can be any column in an existing data frame or a new data frame T he default start column is 1 meaning the data are imported into the specified data frame starting at the first column Preview Only Choose Preview O nly to have the data read entirely as character columns into a data frame called PREVIEW This lets you view all of your data regardless of the data types When you a
321. heck this to display a residual fit spread plot T his is a visual analog of the multiple R squared statistic It compares the spread of the fitted values to the 437 CHAPTER 18 FITTING STATISTICAL MODELS 438 Options Save spread of the residuals Cond Plots of Fitted vs Predictors Check this to display the conditional plots of fitted values versus predictors Include Smooth Check this to display a smooth curve Include Rugplot Check this to display a rugplot on the Residuals vs Fit Sqrt Abs Residuals vs Fit and Response vs Fit plots A rugplot is a sequence of vertical bars along the x axis that mark the observed x values Number of Extreme Points to Identify Enter the number of extreme points that will be identified on the Residuals vs Fit Sqrt Abs Residuals vs Fit Residuals Normal QQ and Cook s Distance plots The row names from the data frame specified on the model page will be used to identify the points Local Regist Model Opiore Regue Pine Mow Data To tl San in VC Predetons F isisi Eros New Data Enter the name of a data frame to use for computing predictions It must contain the same names as the terms in the right side of the formula for the model If omitted the original data are used for computing predictions Save In Enter the name of an S PLUS data frame in which a part such as predictions LOCAL REGRESSION and residuals of the analysis is saved If an object with the na
322. hen you create new Graph sheets T his is convenient if you intend to create many Graph sheets with similar properties To save the settings from the active Graph sheet as the defaults 1 Load a Graph sheet or create a Graph sheet with the desired formatting specifications 2 Click in a blank area of the Graph sheet outside of any graphs 231 CHAPTER 10 FORMATTING A GRAPH 3 From the Standard menu choose O ptions 4 Choose Save Properties as D efault The settings from the current Graph sheet including window size and position are saved as the default settings You can also right click on the Graph sheet and choose the Save Properties as D efault menu option 232 FORMATTING A GRAPH FORMATTING A GRAPH Formattingthe You can select the graph area to change its size or formatting Graph Area Figure 10 1 The graph area gray is selected To select the graph area w Click anywhere outside the axes but inside the graph boundary W hen selected green knobs appear on all sides and all four corners To change the size of the graph area w Select the graph area and drag the corner resize knobs to the desired size or w doubleclick inside the graph area but not on another object to display the Graph Properties dialog Alternatively you can right click and select Position Size from the shortcut menu To change the graph area size specify the H eight and Width on the Position Size page and chooseO
323. hes until you see the panels appear as individual objects 3 Doubleclick on the panel you wish to edit to open its property dialog M ake the desired changes and click OK You can specify standard fill attributes fill color highlight color fill pattern and pattern color and standard line attributes style color and weight for the panel area on the Fill Border page of the Graph Properties dialog TheGraph Properties dialog the M ultipanel page has the following options Panel Type Multipanel graphs allow you to view relationships between different variables in you data set In S PLUS the multipanel feature can be used for standard Trellis graphics to view related data series in a series of panels 2D graphs only or to view a 3D plot at a series of view angles 3D graphs only Column List Specify the column s to use as conditioning variable s D ata Set Specify the data set containing the conditioning variable Type Choose from Auto Discrete or Continuous If the Discrete one panel will be created for each value If Continuous is selected the number of panels is set to the value in the Number of Panel field If Auto is specified S PLUS will try to determine an appropriate setting Order Type Determines the order of your panels Specify Data to order panels in the order that the values appear in the conditioning column Choose Alphabetical to arrange them in alphabetic order of column names CREATING TRELLIS GRAPHS
324. hich summary statistics will be generated To generate summary statistics for all columns select All Variables M aking no selection has the same effect gt Tip You can type into the D ata Frame edit box the name of another S PLUS object such asa vector or a matrix In the case of a matrix the summary statistics will be computed for the columns Statistics Minimum 362 SUMMARY STATISTICS Summaries by Group Check this to generate the minimum value for each numeric column of the data frame First Quantile Check this to generate the first quantile value for each numeric column of the data frame Mean Check this to generate the mean value for each numeric column of the data frame Median Check this to generate the median value for each numeric column of the data frame Third Quantile Check this to generate the third quantile value for each numeric column of the data frame Maximum Check this to generate the maximum value for each numeric column of the data frame Number of Rows Check this to generate the number of rows value for each numeric column of the data frame Number of Missing Rows Check this to generate the number of missing values N As in each numeric column of the data frame Variance Check this to generate the variance estimate for each numeric column of the data frame Std Deviation Check this to generate the standard deviation value for each numeric column of the data
325. hs Creating a Graph Sheet O pening an Existing Graph Sheet Methods for Creating a Graph Viewing Graph Sheets Adding a Plot to a G raph H iding Unhiding and D eleting a Plot Editing D ata Specifications Adding or Replacing D ata in the Plot Changing Columns and D ata Sets Placing M ultiple Graphs on a Graph Sheet Adding aGraph to an Existing Graph Sheet Combining Graphs from M ultiple Graph Sheets Arranging Graphs on the Graph Sheet Preparing D ata for Graphing Projecting a2D Plot onto a3D Plane Combining M ultiple 2D Plotsin 3D Space Combining 2D and 3D Plotson OneGraph Brush and Spin H ighlighting and D ownlighting Points Brush Symbols and Size C ontrol M ore Brushing O ptions Spinning D ata Chapter 9 Working with Trellis Graphics Creating Trellis Graphs Plots on T relis Graphs Formatting Plots within Panels Formating Panels Formatting Panel Strips Trellis Examples viii 162 164 166 168 168 170 173 174 175 175 177 182 183 186 187 187 188 190 190 191 192 194 201 201 203 204 204 205 205 206 207 209 210 212 212 214 216 CONTENTS References Chapter 10 Formatting a Graph Formatting a Graph Sheet Formatting a Graph Formatting the Graph Area Formatting the Plot Area Formatting Panels Formatting 2D Axes Formatting 2D Axis Labels Formatting and Rotating 3D Axes Formatting 3D Axes Labels Formatting 3D Planes Rotating a3D Graph Displaying 3D Multipanel Graphs Displaying 3D Sliced Graph
326. ialog 1 Click in the window or dialog to make it active 2 Click and drag the title bar until the window or dialog is in the desired location To resize a window 1 Click in the window to make it active 2 Position the mouse over one of the four window borders 3 Themouse will change to a double headed arrow when it is over the border 4 Click and drag the border to the desired size To expand a window to maximum size 1 Click in the window to make it active 2 Click the maximize button on the title bar or double click the title bar T he maximize button will change to the restore button To save defaults for a window Right click the mouse inside the window and choose the Save as D efault option from the shortcut menu You can specify window size defaults independently for the Object Browser Graph sheets Data windows Commands window Script windows and Report windows 33 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE Arranging Icons Viewing Multiple Windows Closing Windows 34 This option provides a convenient way to organize your minimized document icons automatically when using Windows 3 1 To arrange minimized document icons w From the Window menu choose Arrange Icons All minimized document icons are arranged automatically on the screen In S PLUS each different type of object such as a Graph sheet data set or script is displayed in a separate window You can also have multip
327. icating how many shades should be used between each core color For example if there are three core colors black red and white and number of shades is specified as 5 15 a total of 23 colors will be used for the image color scheme black 5 shades between black and red 15 shades between red and white Edit Colors Click on the Edit Colors button to access and edit a color palette of the core image colors Only the number of colors specified by the of Colors prompt will be used in the image color scheme In S PLUS you can choose to have plots redrawn automatically after each change made to the graph You may want to turn this feature off to save redraw time for computationally intensive or complicated plots To redraw plots automatically w From the View menu select Auto Plot Redraw To redraw plots manually 1 From the View menu deselect Auto Plot Redraw 2 When you want to redraw the plots choose Redraw Plots N ow from the View menu INDEX INDEX Numerics 2D Plot Palette 178 3D bar charts 335 3D bars 335 3D Plot Palette 179 3D spline plots 335 A Access files 128 adding graphs using drag and drop 191 using Insert Graph 191 add on modules 20 alternative hypothesis 380 382 384 389 400 403 analysis of variance one way 396 analysis of variance table 394 416 442 angle vector plots 328 angle units vector plots 328 Annotation toolbar 277 area charts additive y values 304 by area 304 co
328. ick Position Specify the orientation of the tick marks on the panels Specify whether to have ticks drawn from the axis In from the axis O ut to have ticks Cross the axis or not drawn at all O ff This option is used only when you havea multipanel graph Axis breaks allow you to remove selected portions of your axis and to scale portions of your axis differently This is useful in displaying outlying data pointsin your plot without sacrificing detail in the rest of your graph Broken axes can be created by using options on the Breaks page of the Axis dialog Break Axis C hoose whether to break the axis Start Values Specify a starting value For multiple breaks specify a list delimited by commas End Values Specify an ending value For multiple breaks specify a list delimited by commas Positions axis Specify where to position the break along the axis as a percentage of the total axis length 100 For multiple breaks specify a list delineated by commas 243 CHAPTER 10 FORMATTING A GRAPH Formatting 2D Axis Labels 244 Post Intervals Specify the interval to place labels and ticks after each axis break For multiple breaks specify a list delineated by commas Auto can be used to automatically scale intervals Post First Ticks Specify the value to place the first tick after each axis break For multiple breaks specify alist delineated by commas Auto can be used to automatically calculate first ticks Post Scalin
329. ick on Response to assign the selection as the response Click M ain Effect to assign the selection as a main effect Click on Interaction to assign the selection as an interaction term Click on MainHnteract to assign the selection as main effects and interaction terms Click on Quadratic to assign the selection as a quadratic item Click on Cubic to assign the selection as a cubic item After selecting variables from Choose Variable s click on the list box Term Category to assign the selection as a special term For example all cross terms give you all possible main effects and interaction terms to an order Some special terms have options listed in the box O ption T he syntax of the default option is shown in box Format Click on button Add to enter the term in the formula Check Remove Intercept if you don t want an intercept in the formula W hen predictors are added to the formula they are listed as options in the box Term If you decide to delete some terms select the term first then click on Remove The term is removed from the formula and the option list in Tem Formula The wide string box Formula shows the formula you are building You may edit the box directly or right click on the box and then click on zoom for further editing 537 CHAPTER 20 BUILDING FORMULAS TRANSFORMATION Use the Transformation button in the Formula page to open the Transformation dialog Times oemalion Pilg Yanshle Fungon Thoaia warabba
330. ictors Check this to use the standard normalized values of numeric predictors as predictors in the model Span Select or enter anumber between 0 and 1 to set the smoothness parameter A larger fraction generates a smoother curve No of Par Enter a number analogous to the number of parameters in the modal to specify smoothness Family Choose either gaussian or symmetric as the distribution for the error tem Add Quadratic Term in Local Fitting Check this to use locally quadratic fitting Uncheck this to use locally linear fitting Drop Quad Term s Select the variables for which locally linear fitting will be used Cond Parametric in Select the associated predictors in the case that a portion of the model is of parametric approach LOCAL REGRESSION Save Model Object Formula Options Page Control Parameters Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to aname that starts with last For example last loess is the most recent local regression model fit Formula Enter a formula specifying th
331. ides an essential suite of tools designed for univariate and multivariateG ARCH modeling of financial time series data S SPATIALSTATS provides a comprehensive set of tools for statistical analysis of spatial data including tools for hexagonal binning variogram estimation and kriging autoregressive and moving average modeling and testing for spatial randomness S WAVELETS offers a visual data analysis approach to a whole range of Signal processing techniques such as wavelet packets local cosine analysis and matching pursuits StatLib is a system for distributing statistical software data sets and information by electronic mail FTP and the World W ide Web It contains a wealth of user contributed S PLUS functions e To access StatLib by FTP open a connection to lib stat cmu edu Login aS anonymous and send your e mail address as your password The FAQ frequently asked questions is in S FAQ or in HTML format at http www stat math ethz ch S FAQ To access StatLib with a web browser visit http lib stat cmu edu To access StatLib by e mail send the message send index from S to statlib lib stat emu edu You can then request any item in StatLib with the request send item from S where item is the name of the item S news is an electronic mailing list by which S PLUS users can ask questions and share information with other users To get on this list send the message subscribe to S news request utstat toronto edu To get of
332. iedman For Loess For Spline For Kernel 322 Robust uses the tsreg function to perform a least trimmed squares robust regression T his regression estimate minimizes the sum of the smallest half of the squared residuals Kernel uses the ks moot h function to perform a kernel smooth which is a generalization of local average smoothing Loess uses the oess function to fit a local regression Spline uses thes mooth spline function and thepredict smooth spline function to calculate predictions from a cubic B spline T he regression is fit by penalized least squares between knots For small data vectors n lt 50 a knot is placed at every distinct point For larger data sets the number of knots is chosen judiciously in order to keep the computation time manageable Friedman s Super uses the sups mu function to compute Friedman s variable span smoother It uses a symmetric k nearest neighbor linear least squares fitting procedure The algorithm is fast and by default uses cross validation to pick the span T his allows the user to specify a smoothing function Number of Output Points Specify the number of points to be produced by the smoothing If Auto is selected the number of output points is set to the maximum of 100 and the length of the input vectors Span Fraction of observations in the smoothing window If Auto is selected then automatic variable span selection is done by means of cross validation Reasonable span valu
333. ies Vector 298 fe me lt J P J P J P J P J P J J P P P P P J P j Plot type can be plotted on the respective axes type P A projection of the 2D plot into 3D space PLOT PROPERTIES PLOT PROPERTIES Formatting Plot You can change the properties of your plot using a dialog or the toolbar Properties buttons You can specify line border and fill attributes data handling information cropping information and much more To change plot properties using a dialog The title of the plot property dialog varies depending on the plot type for example Line Scatter for line and scatter plots You can access the plot property dialog in the following ways 1 5 D ouble click on the plot or 2 Select a plot and choose Selected Plot from the Format menu or 3 4 Expand the O bject Browser object GraphSheet until you see the plot Right click on the plot to display its shortcut menu or icon Double click on the icon or right click and select Property Dialog from the shortcut menu Edit the desired properties in the dialog and choose O K To change plot properties using the toolbar 1 2 Select the plot to change From the Graph toolbar select the desired button for example Color Choose a new color for the plot Note If you have a line plot with symbols both the lines and symbols will change color You
334. if you wish to edit it again in S PLUS Note In order to edit or manipulate traditional graphs created using the Commands window or the Statistics menus and dialogs you must first convert these graphs to object oriented graphs This is done by right clicking on any line point or text in the graph then selecting C onvert to O bjects 174 S PLUS GRAPHS Table 8 1 The Graph toolbar Arial Set font Set text size Bold text Send to back Fill color Auto legend Creating a To create a Graph sheet Graph Sheet Italic text Bring to front Line Symbol color Annotations palette Underline Superscript Pattern Line style Send graph to other application Subscript Line weight 1 Click the N ew button or choose N ew from the File menu 2 Select GraphSheet from the list of available document types 3 ChooseOK to create the Graph sheet Opening an Graph sheets can be saved as files and then opened again at any time Existing Graph Sheet To open an existing Graph sheet 1 Click the O pen button or choose O pen from the File menu S PLUS lists the files in the current directory 175 CHAPTER 8 CREATING A GRAPH 2 In the File Name box type the file name or select the file you want to open 3 To list other files you can change the drive or directory or select the specific file type in the List Files of Type box T his displays a list of all files of one t
335. ifications 316 error calculations 316 horizontal bars 316 plot options 316 user defined 316 vertical bars 316 error bars automatic 306 bar charts 305 ethanol data frame 70 216 exact binomial test 400 exact distribution 382 example data sets 55 example O bject Browser 56 examples Commands window 81 environmental 78 ethanol 70 exenvirn 65 fuel frame 57 galaxy 61 Powerpoint presentation 83 examples O bject Browser 146 Excel files 127 exenvirn data frame 65 explode amount pie slices 326 Export D ata dialog D ata Specs page 136 O ptions page 137 exporting data 135 exporting column names 137 exporting specifying line length 137 F factor analysis 493 FASCII files 128 FASCII importing specifying a format string 128 files importing 117 fill color 302 color surface plots 333 direction area charts 304 method surface plots 334 pattern 302 type plot properties 302 fill color styles 603 Fill Expression field 102 fills shading and colors 333 filtering in the O bject Browser 143 filtering options 151 Find button 553 Find S Plus O bjects option 146 first quantile value 363 Fisher s exact test 404 fixed effects analysis of variance 444 fixed effects linear model 480 folders 153 formats numeric columns 114 formatting columns 112 data sheets 111 formula builders 535 formula building example 78 formula dialog 535 formulas Add group box 537 Cox proportional haxards 541 H eart data frame example 543 line
336. ify a beginning and ending x y position for each vector Specify a column of x starting values under X a column of y starting values under Y a column of x ending values under Z and acolumn of y ending values under W 327 CHAPTER 12 EDITING PLOT PROPERTIES Scatterplot Matrix Data to Plot Line Histogram Symbol and Vary Symbol and Smoothing Matrix Plot Axes 328 If you choose Angle M agnitude you need to specify an x y position for the vector the angle at which to draw the vector and the magnitude or length of each vector Specify the x data under X the y data under Y a column of angle data under Z and acolumn of magnitude data under W Vector Position Choose whether the x y data should be used to define the M idpoint H ead or Tail of the vectors Angle Units Specify whether the angle data should be interpreted in D egrees or Radians Magnitude Multiplier Specify a value to proportionally increase or decrease the length of the vectors For example if you specify a value of 0 5 the vectors are drawn at half their original magnitude or length The default value is 1 which means the magnitude data determines the length of the vectors A scatterplot matrix is an array of pairwise scatter plots showing the relationship between any pair of variables in a multivariate data set To customize the layout of your scatterplot matrix see section Matrix Graph page 329 To customize the axes and labeling see section
337. ight all rows equally leave this blank Use Cross Validation to Set Span Check this to specify that cross validation is to be used to automatically select the variable spans used in the smoothing Uncheck this box to allow setting of the Smoothing Parameter Smoothing Span Select a numeric value between 0 and 1 here to bypass automatic span selection T his field is enabled only when Use Cross Validation to Set Span is unchecked Bass Frequency Enter a numeric value here to control the low frequency emphasis when using cross validation The larger the value up to 10 the smoother the fit from automatic span selection Values less than O or greater than 10 are essentially the same as 0 Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This object will be saved as alist with components named x and y This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last lm is the most recent linear regression model fit Plot Smooth Check this to display a scatterplot with the smoothed curve Related S PLus language functions supsmu
338. ile the windows Select Type from the right pane of the O bject Browser and drag it to the top part of the graph A rectangular drop target will appear D rop it inside of the rectangle The graph will be drawn with a different panel for each automobile type 5 Change the line style of the crosshairs using the toolbar Click on any crosshair or symbol to select the plot Click on the Line Style button on the graph toolbar Select the dashed line T he crosshairs will be redrawn in each panel using a dashed line Redrawing the graph with different variables 6 Select Weight then cTRL click to select Fuel from the Object Browser 7 Drag the selected columns over to the graph As you hover over a point on the scatter plot the points will change color D rop the data on the scatter plot T he plot will be recalculated and redrawn using the Weight and Fuel Consumption data The plot should now look 225 CHAPTER 9 WORKING WITH TRELLIS GRAPHICS similar to that in figure 9 8 Figure9 8 Theresulting Trelis plot for Example 4 References Becker R A Cleveland W S and Shyu M 1996 The Visual Design and Control of Trellis Display Journal of Computational and Graphical Statistics 6 pp 123 155 Cleveland William S 1993 Visualizing Data ATT Bell Laboratories Murray Hill NJ 226 FORMATTING A GRAPH Formatting a Graph Sheet Formatting a Graph Formatting the Graph Area Formatting the Plot Are
339. in the Import to group are ignored for files of type S PLUS Transport 122 FILTER PAGE FILTER PAGE File gama Diets Boeci 7 Options Files Bie brigrresbr Files ol yee Ted GAS fh Fide f pon be ae import To Cata Game Tega San Got i z Erea e Figure5 2 TheData Import dialog Filte page The Filter page allows you to subset the data you import By specifying a query or filter you gain additional functionality such as taking a random sampling of the data U sethe following examples and explanation of the filter syntax to create your statement A blank filter is the default and results in all data being imported Note The Filter Page for the File Import dialog is ignored if the File of Type field is set to ASCII formatted ASCII and S PLus Transport Case Selection You can select cases by entering a caseselection statement in the Filter Information box in the Filter dialog T he case selection or where statement has the following form where variable expression relational operator condition 123 CHAPTER 5 IMPORTING AND EXPORTING DATA 124 Variable Expressions You can specify a single variable or an expression involving several variables All of the usual arithmetic operators are available for use in variable expressions Relational Operators The following relational operators are available O perator equals l not equal
340. including M ixed Decimal Scientific Currency Financial Date Date amp Time Time and Elapsed Time H M S To change the display format of numeric data 1 Doubleclick in the column header or select the column and choose Selected O bject from the Format menu 2 Choose a new display format for example Currency 3 ChooseOK You can change the precision for anumeric column by specifying the number of digits to be displayed after the decimal The maximum number allowed is seventeen T his only affects the way numbers are displayed and has no effect on internal computations which always use the maximum available precision FORMATTING DATA WINDOWS To Set Column Defaults To change column display precision 2 3 0 To increase the precision click on the Increase Precision button until you see the desired precision 00 To decrease precision click on the D ecrease Precision button 3 or Doubleclick in the column header or select the column and choose Selected Column from the Format menu Type in the desired number in the Precision field ChooseOK You can set column defaults justification precision width etc You could for example have a different default width for numeric columns than for character columns To set column defaults 1 Double click in the column header or choose Selected O bject from the Format menu Specify the desired settings in the dialog Right click in the colum
341. ines at the Means 224 8 Under Customization set Skip Column to c F F F F T F F F F The fifth panel on the graph will be skipped when drawing into the panels Panels are counted beginning with the lower left hand corner because the default order is set to Graph Order Under Customization set Panel Order to c 1 2 3 5 8 7 6 4 This will rearrange the order of the contents of the panels actually used For example what would normally be drawn in the fourth panel middle row left column will now be drawn in the last panel top row right column Click OK Add annotation to the middle panel 8 Open the Annotation palette and click on the Comment tool Left click on the position in the middle panel where you would like to start your text then drag up and right until you have defined the approximate size of your text Click on the selection button on the Annotation tool then click on the new text Edit the contents to say Velocity of Galaxy NGC 7531 Click outside of the text to end the editing Changing the plot type Depth perception is typically better with a 3D mesh than with a 3D scatter plot Let s change the plot type to a surface plot to further examine the data 9 Click on any of the points in the scatter plot to select it N ow click on the Data Grid Surface plot icon on the 3D plot palette The 3D scatter plot will be changed to a mesh plot Because the data are not gridded i e the data points are a series of triple
342. ing menu am singer kafiz environmental exsurf fuel frame kyphosis last dump rs tmp Figure21 1 A Script window howing the program pane top and output pane bdow THE SCRIPT WINDOW THE SCRIPT WINDOW Any executable statements or commands in S PLUS can be entered into a Script window and executed For example you could enter the following expression in a Script window obj ects All Script windows have an output pane which contains output from print statements and information about warning and error messages that occur when running your script The Script window program pane contains a line and column number indicator in the upper left of the window that helps you locate linesin your script when you are editing Table 21 1 The Script toolbar Scr pt Run selected Find text script Working with Scripts can be created or opened edited run saved and printed Scripts To create a new script 1 Click the N ew File button or choose N ew from the File menu A list of window types will pop up 547 CHAPTER 21 WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS 548 2 Select Script File and click OK New Object Browser Report File Graph Sheet Data Frame OK Cancel rth Help Figure21 2 Open N ew can be used to create many file types including a Script file A new script is created and displayed in a window N ew scripts are given t
343. ing M ultiple O bjects 291 Arranging O bjects on the Graph 292 Overlapping O bjects 292 Aligning O bjects 292 Aligning O bjects Using Snap to Grid 293 Distributing O bjects 293 Graphic All objects on a Graph sheet are considered graph objects W hen working with Objects graph objects you can Change the shape size and color of objects Layer objects Copy and move objects Align objects Once an object has been added to your graph you can move copy size or edit the properties of the object 281 CHAPTER 11 WORKING WITH GRAPH OBJECTS Selecting Objects 282 You can also use the O bject Browser to select edit and format objects The O bject Browser allows you to see all of the objects in your current session in a hierarchical view It can also be used to select and edit objects that are difficult to select in other views for example overlapping objects For more information see chapter 6 U sing the O bject Browser To select an object the mouse pointer must appear as an arrow If the pointer isnot an arrow click on the Selection tool on the Annotation palette W hen a drawing object is selected small green selection knobs appear on an outline around the object For a box the outline is the same as the border around the box for arcs ellipses and extra symbols the outline is an invisible rectangle surrounding the object For lines radial lines and arrows the green selection Knobs appear only at the ends
344. ing it on the left pane of the target Object Browser A Browser Page can also be dragged onto the toolbar and dropped to create a toolbar button that will recreate the Browser Page To modify the filter on an existing Browser Page select the Format Browser Page menu item or right click in the white space of the left pane of the O bject Browser and select the Filter menu item from the context menu The filtering options found on the Filter tab of the Browser Page dialog are outlined below An exampleis shown in Figure 6 5 Show Only Folders Check this to display only folders and their contents in the O bject browser O ther objects will not be displayed Interface C lasses Select the interface classes which should be displayed in the Object Browser page Interface classes include the following Automation Client Classinfo FunctionInfo GraphSheet M enultem O bjectD efault Property Script SearchPath and Toolbar Use the CTRL key to make multiple selections or the shift key to make group selections D atabases Select the database directories or attached objects which should be searched for objects to display in the Object Browser page Only objects 151 CHAPTER 6 USING THE OBJECT BROWSER 152 which are found in these locations will be displayed Classes Select the S PLUS object classes which should be displayed Use the CTRL or shift keys to make multiple or group selections User defined classes may be specified by typing th
345. ing language equivalents of these operations are recorded in the H istory log You can view the History log in a Script window In order to record dialog operations in the H istory log you must use the OK or Apply buttons in the dialog to accept your changes If you choose C ancel or press ESC from the dialog the command which corresponds to the dialog isnot recorded in the H istory log You can edit lines in the H istory log just as you would any other script T hese edits do not modify the H istory log itself they only modify the copy in this Script window You can cut and paste parts of the script into other scripts execute portions of it or the entire script and save the script to a file The maximum size of the History log the total number of operations recorded can be specified in the History Entries field of the Undo and H istory dialog available through the O ptions menu You should clear the History log before you start recording steps T his will save editing time later and make it clearer what commands were generated by the actions you made in menus and dialogs To view the current History log in a Script window 1 Click the History Log button to display the H istory log with default TIME SAVING TIPS FOR USING SCRIPTS IN S PLUS settings or from the Window menu choose H istory then Display The Display H istory Log dialog appears 2 Specify any desired display options 3 ChooseOK to display the H istory log
346. ing the desired model Create Formula Click this to open a formula builder dialog used to construct a formula specifying the desired model See the chapter on Building Formulas for more information Curve Type Select a type of survival curve kaplan meier flening harrington or fh2 T he default is kaplan meier Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to aname that starts with last For example last survfit is the most recent nonparametric survival model fit 507 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS Model Optors Rosie Plot Conhdence Lime F Compute Standen Enos Sudan Eso Porras qreervood Cosi desce interval Typa is Lower irrita e1 Options Page Confidence Limits Compute Standard Errors 508 Check this to compute standard errors Confidence Interval Level Enter anumber between 0 and 1 to specify the confidence interval level Standard Error Formula Select the error formula greenwood for the Greenwood formula or tsiatis for the Tsiatis form
347. ion Hide Choose whether to hide or display the text If you want to unhide the text once you have closed the Comment dialog you can reopen the dialog from the O bject Browser Use Axes Units Choose whether the position specification should be interpreted in document units or axes units X Axis and Y Axis f you choose Use Axes Units these options become available If you have more than one x or y axis on the graph these options let you specify which x axis and which y axis you want used for determining the axes units Draw in Panels For a multipanel graph specify the panels in which you would like your text to appear On the Font page you can specify standard font attributes for the text 263 CHAPTER 10 FORMATTING A GRAPH Box Page 264 Specifications made on the Font page apply to the entire specified text On the Box page you can specify standard line and fill attributes for the box around the text You can also specify the following options Vertical and Horizontal Margin Specify the distance between the box border and the text ADDING SPECIAL CHARACTERS AND FORMATTING TEXT ADDING SPECIAL CHARACTERS AND FORMATTING TEXT Specifying a Text To specify a text font Font 1 Open atext edit box and add text 2 Highlight the text to be formatted 3 Choosea font from the toolbar s font dropdown list Specifying Text To specify a text size Size 1 Open atext edit box and add text 2 Highlight the tex
348. ion about a particular action In S PLUS there are two types of dialogs action dialogs and property dialogs Dialogs Action dialogs carry out commands such as copying a column Property dialogs display and allow you to modify the properties characteristics of the selected object Dialogs can contain multiple tabbed pages of options To see the options on a different page of the dialog click the page name or press cTRL TAB to Move from page to page When you choose OK or Apply or press ENTER any changes made on any of the tabbed pages are applied to the selected object M ost of S PLUS s dialogs are modeless They can be moved around on the screen and they remain open until you choose to close them T his means you can make changes in a dialog and see the effect without closing the dialog This is useful when you are experimenting with changes to an object and want to see the effect of each change T he Apply button or cTRL ENTER can be used to apply changes without closing the dialog W hen you are ready to close the dialog you can choose Cancel press ENTER or ESC or just double click the Close box on the dialog Note Choosing OK will close the dialog and execute the command specified in the Dialog If you do not wish the command to execute after the dialog closes perhaps because you have already clicked on Apply choose Cancel instead of OK The OK Cancel When you are finished setting options in a dialog box
349. ions and will only evaluate then on demand The script can be executed by clicking on the Run button If a section of the script is selected highlighted only that selection will be executed T he output is shown in the output pane and not below each command The Commands window is preferable for the user doing interactive exploratory data analysis at the prompt while the Script window is useful for writing longer functions sg 1 546 maxit bf maxit bf maxit trace trace fun call control lt control list fun args lt is element arg names fun call arg names gam fun call lt fun call c T fun args gamobj lt eval fun call summary function tabSummary gam gamobj print short p print long p print anova p Save name save fit p save resid working p save resid pearson p save resid deviance p save resid response p plot function if any c plotResidVsFit p plotSqrtAbsResid p plotResponseVsFit p plotQQ p plotPartialResid p tabPlot gam gamobj predict plotResidVsFit p plotSqrtAbsResid p plotResponseVsFit p plotQQ p smooths p rugplot p id n plotPartialResid p plotPartialFit p rugplotPartialResid p scalePartialResid p if any c predict p se p tabPredict gam gamobj newdata predobj name predict type predict p se p invisible gamobj wor Last fixed Last value a as character or error beer catalyst ati ethanol guayule last warn
350. ions kept in the working database Alternatively you may filter for object classes in a large S PLUS database or multiple databases Filtering for objects is done through the Browser Page 143 CHAPTER 6 USING THE OBJECT BROWSER The Browser Page Versus the Object Browser Organization of Objects 144 T heleft pane of the O bject Browser displays the active Browser Page A single O bject Browser window can have multiple Browser Pages each with its own filter specification and is associated with a tab located on the bottom left hand side of the window frame The Browser Page and the Object Browser window both have distinct dialogs The Object Browser dialog contains the properties specific to the Object Browser and display characteristics of the right pane The Browser Page dialog contains properties specific to the filtering characteristics of that page Both the Object Browser and the Browser Page have a context menu the O bject Browser context menu is accessible by right clicking on the white space of the right pane and the active Browser Pages context menu is accessible by right clicking on the white space of the left pane T he O bject Browser context menu contains menu items for creating a Browser Page saving the Object Browser closing the O bject Browser and access to its properties dialog The Browser Page context menu contains menu items for creating a Browser Page inserting a folder into the page or into the selec
351. iptions displayed on the status bar when you pause the mouse over a dialog field Show Commit Dialog on Exit Choose this option to have S PLUS always prompt you to save or cancel changes made upon ending an S PLUS session Open at Startup Choose whether to have S PLUS always open the O bject Browser the Commands window or both at startup System Debug Mode If checked S PLUS performs various internal checks during evaluation This provides more information about warning messages and reloading and may help track down mysterious bugs such as S PLUS terminating abnormally Evaluation will be substantially slower with this option turned on and at times introduces strange behavior Error Action Specify the function with no arguments to be called when an error or interrupt occurs S PLUS providesdump calls and dump frames to dump the outstanding function calls or the entire associated frames See the on line documentation for these functions for details Setting the function to NULL eliminates all error actions Warning Action Specify the level of warnings that you would like reported no warnings collect warnings and report them at the end of the evaluation report warnings as they occur or convert warnings into errors to terminate the evaluation of the expression Max Recursions Specify the maximum depth to which expressions can be nested T his exists primarily to catch runaway recursive calls of a function to itself directly or in
352. is 373 M magnitude vector plots 328 M antel H aenszel chi square test 408 margins comment plots 310 pie charts 325 matrix 88 maximum value 363 M cN emar s chi square test 406 mean value 363 median value 363 menus adding a plot with 185 merging data frames 351 modifying data 155 modules add on 20 multipanel graphs data format 199 multiple comparisons 480 multivariate analysis of variance 488 naming columns formatting columns See new data frame matrix or vector 348 N ew Toolbar dialog 568 nonlinear cf plots 324 nonlinear regression 421 nonparametric survival modeling 506 normal approximation 382 number of decimal places 115 611 INDEX numeric column formats 114 0 O bject Browser 141 configuration files 148 copying data 156 default 146 default and example 56 deleting objects 156 dialog 148 examples 146 filtering 143 filtering options 151 folders 153 left and right panes 142 modifying data 155 object creation 155 shortcut keys 147 toolbar 142 tutorial 65 object defaults modifying 598 objects saving defaults 598 ODBC configuring a data source 132 exporting files 138 files and tables 130 installation 130 offsetting pie chart labels 326 text in comment plots 310 OLE objects 162 omit constant curve fitting 314 one sample chi square test 386 one sample K olmogorov Smirnov test 384 onesamplet test 380 612 one sample Wilcoxon test 382 one way analysis of variance 394 on lineh
353. is C ovariance M atrix Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case 501 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS Results Page Printed Results Loading Options 502 sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last princomp is the most recent principal components analysis modal fit Include Scores Check this to have the factor scores returned as a component of the fitted model Poncipal DHT pipra rig Annahme r i z Bll Moda Feau Pint Predici Froris Denie Loadnjs Options FF Shota P Comen meotan r Laingia Short Output Check this to print a summary of the model results in the designated output window Printed results include sums of squares of the component loadings the size of the data the names of the components in the fitted model object and the call that created the model object Component Importance Check this to include the importance of each component in the printed results Loadings Check this to include the loadings matrix with the printed results Cu
354. is example creates a formula to model survival as a function of age and extent of residual disease residual dz The variables futime and fustat contain the time and event information respectively In the formula field the survival response appears as a call to the Surv function The arguments to Surv are the variables selected for Time 1 Time 2 if applicable Censoring Indicator and Origin if applicable The variable selected as the Censoring Indicator call can be coded using 0 alive 1 dead Other choices include 1 2 or T F where T or 2 indicates death For interval censored data Parametric Survival only 0 right censored 1 event death 2 left censored and 3 interval censored Seethehelp fileon Surv for more details 1 Choose Statistics Survival C ox Proportional H azards from the main menu 2 In the D ata Frame field enter or select ovarian then click the C reate Formula button 3 Create the Survival Response by selecting futime from the Choose Variable s list and click the Time 1 button T hen select fustat from COX PROPORTIONAL HAZARDS the Choose Variable s list and click the C ensoring Indicator button Click the Add Response button 4 Define explanatory terms by holding down the CTRL key and simultaneously select age and residual dz from the Choose Variable s list Then click the main effects button from the Add Term s group 5 A basic survival formula is now created Click the OK button to acce
355. is to print the results in the designated output window Related S PLus language functions aov im menuOneway 395 CHAPTER 17 COMPARING SAMPLES KRUSKAL WALLIS RANK SUM TEST 396 Data Results T his dialog performs a Kruskal Wallis rank sum test on data following a one way layout This is a non parametric alternative to a one way analysis of variance See the on line help for kruskal test for details on the form of the statistic used and assumptions of the test To perform a Kruskal Wallis rank sum test Choose Statistics gt Compare Samples gt k Samples gt Kruskal Wallis rank sum from the main menu T he dialog shown below appears Kreskat Wrallie Bank Som est Dota Faria Dieis Frene See AT Fasgain ariana F Pmi Gesets Grouping engin Data Frame Select a data frame Response Variable Select the column that contains the response variable T his must be numeric Grouping Variable Select the column that indicates group membership for the response above T his is usually a factor with k levels Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten The saved object will have classht est This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period
356. isk to denote a missing name e g Apples Pears Format String Specify the data types of the imported columns For each column you need to specify a sign and then the data type Dates may automatically be imported as numbers After importing you can change the column format type to a dates format O ne of the delimiters specified in the D elimiters field must separate each specification in the string Here is an example ASCII format string ws Wf f The s denotes a string data type f denotes a float data type and the asterisk denotes a skipped column If you do not specify the data type of each column S PLUS looks at the Start Row and uses the contents of this row to determine the data type of each column A row of data must always end with a new line Note that field width specifications are irrelevant for ASCII files and are ignored M ultiple delimiter characters are not grouped and treated the same asa single delimiter For example if the comma is a delimiter two commas are interpreted as a missing field Double quotes are treated specially They are always treated as an enclosure marker and must always come in pairs Any data contained NOTES ON IMPORTING FILES Notes on Importing Excel Files Notes on Importing Files with Multiple Tables Notes on Importing Lotus Files between double quotes are read as a single unit of character data T hus spaces and commas can be used as d
357. it Spread Check this to display a residual fit spread plot T his is a visual analog of the multiple R squared statistic It compares the spread of the fitted values to the spread of the residuals Cond Plots of Fitted vs Predictors FIXED EFFECTS ANALYSIS OF VARIANCE Options Partial Residual Plots Check this to display the conditional plots of fitted values versus predictors Include Smooth Check this to display a smooth curve computed with loess smooth on the Residuals vs Fit Sqrt Abs Residuals vs Fit and Response vs Fit plots See the on line help for oess smooth for details Include Rugplot Check this to display a rugplot on the Residuals vs Fit Sqrt Abs Residuals vs Fit and Response vs Fit plots A rugplot is a sequence of vertical bars along the x axis that mark the observed x values Number of Extreme Points to Identify Enter the number of extreme points that will be identified on the Residuals vs Fit Sqrt Abs Residuals vs Fit Residuals Normal QQ and Cook s Distance plots The row names from the data frame specified on the model page will be used to identify the points Plot Partial Residuals Check this to display partial residual plots for all the terms in the modal Include Partial Fit Check this to display the partial fit for the term Include Rugplot Check this to display rugplots on the partial residual plots Common Y axis Scale Check this to give all the partial residual plots the same v
358. it button marked by an dli pse A scatter plot matrix is displayed plotting each column of data against the other selected columns For example to see how M ileage and Fuel are related 58 QUICK TOUR read across from M ileage and above Fuel to see the plot T he plot shows that Mileage and Fuel are directly related You can also see a strong relationship between M ileage and Weight heavier cars have lower mileage Weight Mileage owo o amp ie is o fe o o o ocuD ooo ao o o ojooo coco o o Fuel 0 ance o D oompo 00 ooo oo OGM co co 0000 oo 9 00000 oc000 o Type ee re a a ea r 2 5 3 0 3 5 4 0 4 5 5 0 5 5 Figure 3 5 TheScatterplot matrix howsa number of strong rdationships Linear Regression Now that you re familiar with your data let s examine the relationship between Weight and M ileage a bit more extensively of Weight vs Mileage Conditioned by Type 1 Close the window containing the scatter plot matrix so you can see the D ata window again 2 Click on the header of Weight and then CTRL CLICK on the header of M ileage in the D ata window 3 Click on the Linear Fit button see figure 3 4 on the 2D plot palette This linear fit shows an obvious relationship between an increase in Weight and a decrease in M ileage To examine how Vans or Compact cars fit into this example you can use the exclusive Trellis graphics in S PLUS t
359. izontal bar charts Grouped bar chart with error bars For a grouped vertical bar chart with error bars specify the names or column numbers of your X Y and Z columns The X column determines the position of the groups along the x axis Each series of bars for example the first bar in each grouping as you move across the Graph sheet must havea y value corresponding to each x value Thus the total number of y values must be the same number of x values multiplied by the number of bars in each grouping T hese y values can be specified with multiple y columns or with onelong column TheZ column should contain the standard error data and should have the same dimensions as your Y column s Error bars cannot be 305 CHAPTER 12 EDITING PLOT PROPERTIES Position Fills Error Bars 306 automatically calculated for grouped bar charts Error bars cannot be displayed in a stacked bar chart Thex and y data specifications are reversed for horizontal bar charts Direction of bar charts Specify the bar direction as Vertical or H orizontal Bar Grouping Specify Standard Stacked or Grouped Bar Width Specify the width of the bar as a fraction of the X interval Y interval for horizontal bar charts as determined by the number of ticks or intervals For grouped bar charts the bar width refers to the width for the whole group and not each individual bar Bar Offset Grouped Bar Charts Only Specify the amount of offset between the bars T
360. j ect for 415 CHAPTER 18 FITTING STATISTICAL MODELS 416 Results Page Printed Results Saved Results more information about the saved object EEN Fiant Pin Pede Prinia Fau Smved Resis l Shot Sisi in E Long Ougan E TEET r ANOVA Tohis V Fiesa I Cormelshon Wisin of Eshneates Short Output Check here to display a short summary of the model fit This includes the model formula the regression coefficients the residual standard error and the degrees of freedom Long Output Check here to display a detailed summary of the modal fit T his includes the model formula a five number summary of the residuals the coefficients their standard errors t statistics and p values the residual standard error and the degrees of freedom the multiple R Squared value and the F test for the overall moda with its degrees of freedom and p value ANOVA Table Check here to display an analysis of variance table The sums of squares in the table are for the terms added sequentially Type sums of squares Correlation Matrix of Estimates Check here to display the correlation matrix of the regression coefficients This is only available if the Long O utput is selected Save In Enter the name of an S PLUS data frame in which fitted values and residuals of the analysis are to be saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that a
361. ja wegh Ahaire absid Exp mpi ARG a Log kei Legien Typa j i Ey Sq Foni amtii Pecpiaenl a i1 pks aie i we Oubao tubes Furen wh Cona Funcion Figure20 2 The Transformation dialog Tip A list of variables in the formula dialog is required to open the transformation dialog Whe the tranformation dialog isup the modd and the Formula dialogs are disabled until OK or Canca is clicked Select variables select a function then add thenew variable to thelist T henew variableslist will be sent back to the formula dialog after you click on OK Variable Variables listed in the formula dialog are displayed in the multiple selection box Choose Variable s W hen it is appropriate hold down the CTRL key to select more than one variable After variable selection select a function to apply to the selection The transformed variables are displayed in the box New Variable s Click on Add to add the transformed variables to the variable list in Choose Variable s Function Click on a button in this group to apply a transformation to the selected variables The eight buttons perform the transformations Absolute abs x for absolute value Exp exp x for exponential Log log x for natural log Log10 log10 x for log base 10 Sq Root sqrt x for square root Reciprocal X 1 for reciprocal Quadratic x 2 for quadratic Cubic x 3 for cubic 538 TRANSFORMATION Function with Select a
362. k D ataacc directory of the AccessCD To install the ODBC Administrator simply insert the S PLUS CD and select Install ODBC in the dialog that starts up If you are using Windows 3 x Windows For Workgroups or Windows NT 3 51 you will need to run SETUP EXE in the root directory of the setup media to initiate the dialog S PLUS automatically is installed with an ODBC driver that connects to S PLUS With this and the ODBC Administrator you only need a database or other application and the corresponding driver that connects that application to the ODBC interface Consult your database ODBC driver vendor for instructions on how to install their driver and database or other application IMPORTING ODBC TABLES Figure5 3 TheODBC import dialog ODBC page above Filter page bdow 131 CHAPTER 5 IMPORTING AND EXPORTING DATA Importing Tables from an External ODBC Data Source Configuring an ODBC Data Source 132 1 From the File menu choose Import Data then From ODBC Connection 2 From theODBC page select the data source from the D ata Source dropdown list If the desired data source is not available or the list is blank you may create a data source by selecting the button labeled to the right of the O DBC Data Source list box 3 Once you have selected a data source the Table Name drop down list will be initialized and you can select a Table to import 4 Click OK to import the table D etails
363. k this to indicate that D ata Frame represents different time epochs for only one individual When checked only one curve will be produced By default multiple rows indicate multiple individuals and there will be one curve generated per row in D ata Frame Show Confidence Intervals Check this to plot confidence intervals Confidence Interval Level Enter anumber between 0 and 1 to specify the confidence interval level Confidence Interval Type Select a confidence interval type log plain or log log log for intervals based on the cumulative hazard or log survival T his is the default e plain for standard intervals log log for intervals based on the log hazard or log log survival Line Color s Enter alist of integers using either the name of an existing S PLUS vector ora comma delimited list If the number of curves to be plotted is greater than the length of the list the line colors will cycle through the list Line Type s Enter alist of integers using either the name of an existing S PLUS vector ora comma delimited list If the number of curves to be plotted is greater than the length of the list the line types will cycle through the list Line Width s Enter alist of integers using either the name of an existing S PLUS vector ora comma delimited list If the number of curves to be plotted is greater than the length of the list the line widths will cycle through the list Log Axis for Y Check this to plot the y
364. ked Y Data If your y data are stacked the number of areas will be equal to the number of rows in column Y divided by the number of rows in column X The first y value will correspond to the first x the second y to the second x and so on repeating throughout the y column See section Preparing Data for Graphing page 194 for an example of stacked data 303 CHAPTER 12 EDITING PLOT PROPERTIES Options area charts Fills By Area Bar Charts Vertical and Horizontal 2D Data to Plot 304 Fill Direction Choose Under data to fill below the curve to the x axis Choose Above data to fill above the curve to the top edge of the y axis Choose Left of data to fill to the left of the curve to the left y axis Choose Right of data to fill to the right of the curve to the right edge of the x axis Choose Inside data to fill within the area defined by the data Y Values C hoose Additive to have the first row of y data plotted along the x axis with subsequent rows of y data stacked on top of previous rows T he points graphed represent the total of all values below them Choose Individual to have each row of y values plotted independently on top of each other along the x axis For area charts you can choose a solid color for each area fill the area chart with a series of shades determined by a range of colors or specify new colors for the filled areas using the standard color dialog See the information on fills in the s
365. l Totals Check this to display the row and column totals with the contingency table Run Chi Square Test Check this to perform a Chi Square test Decimal Places Enter the number of decimal places with which to display each proportion Related S Plus language functions crosstabs print crosstabs 367 CHAPTER 16 SUMMARIZING AND EXPLORING DATA CORRELATIONS AND COVARIANCES This dialog provides an easy way to compute correlations and covariance matrices for numeric columns of a data frame or matrix To compute correlations and covariances matrices Choose Statistics gt Data Summary gt Correlations from the main menu The dialog shown below appears Corel fice maj eran nee mE x Cats Frame P Waerence Soo e emablos a F neiii Fraction it Tier fo oe Mothodin Handle Wissing auas reefs Jin F Piet ouets Data Data Frame Select a data frame gt Tip 368 You can typeinto the D ata Frame edit box the name of another S PLUS object such asa matrix In that case etimates for the columns will be computed Variables Select the columns for which correlations or covariance estimates will be generated To use all columns either select All Variables or make no selection Method to Handle Missing Values Select from fail omit include and available See the helpfile for the S PLUS function cor for a detailed explanation of each selection Variance Covariance Check this to genera
366. l be overwritten The saved object will have classht est This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names Print Results Check this to print the results in the designated output window Related S PLus language functions prop test print htest menuProp 403 CHAPTER 17 COMPARING SAMPLES FISHER S EXACT TEST 404 Data Results T his dialog tests the independence between the row and column variables of a two dimensional contingency table The table is determined by two Classification variables in a data frame To perform Fisher s exact test Choose Statistics gt Compare Samples gt Counts and Proportions gt Fisher s Exact from the main menu T he dialog shown below appears Data Femina Ba Cents Frome Seve As Clas ioahon veao Osa F Prt Resets lass icstion variae Ta Data Frame Select a data frame Classification Variable One Select the column from the data frame which offers the first classification or grouping variable T his must bea factor or category Classification Variable Two Select the column from the data frame which offers the second classification or grouping variable T his must be a factor or category Save As Enter the name for the object in which to save the result
367. l values for the AR coefficients to be used by the optimizer This must have length equal to p the order of the autoregressive operator T he default is zero for initial values MA coefficients Enter the vector of initial values for the MA coefficients to be used by the optimizer This must have length equal to q the order of the autoregressive operator T he default is zero for initial values Other Predictors Add a Time Series Covariate Check this to include other covariates in the model Time Series Enter the name or expression for a univariate or multivariate time series or a vector or matrix T hese will be used as additive regression variables Results Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he 529 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names T he default is last arima Print Results Check this to print a summary of the fitted model in the designated output window AAA iedeling pE Filira Opinia Coegnonic Facai Menamem Hambar Irak ota i E Likelihood Evade an Fitting Options Page Maximum Iterations Number
368. lace with Replace Replace All M Match whole word only Cancel I Match case Figure21 6 The Replace dialog has the same text length limits as the Find dialog 4 Inthe Replace With box type the replacement text As with the Find W hat box if you used Replace With in your current work session the replacement characters you last specified are selected in the Replace W ith box Type over the text to specify different replacement characters Choose the Find Next to move the cursor to the next occurrence of the word in Find W hat Choose Replace to replace the current occurrence of the word in Find W hat with the word in Replace W ith Choose Replace All to replace all occurrences of the word in Find W hat with the word in Replace W ith with no confirmation dialog You can also delete text with the Replace option Follow the steps above but leave the Replace W ith box blank Hiding and You can hide scripts to reduce the number of windows on your screen Unhiding Hidden scripts can still be accessed through dialogs and other scripts but cannot be edited directly until they are made visible Scripts 554 THE SCRIPT WINDOW To hide a script w From the Window menu choose H ide To unhide a script 1 From the Window menu choose Unhide A list of hidden Script windows will pop up 2 Doubleclick the script you wish to unhide or select one and click OK Cancel Figure21 7 Unhidinga script
369. lder is selected when a new folder is inserted the new folder will be placed inside the first folder Folders are owned by the Browser Page To add an object to a folder simply drag the icon of an object onto the folder or use copy and paste A shortcut to the object is created T his allows data functions and other objects to be conveniently placed in a central location 153 CHAPTER 6 USING THE OBJECT BROWSER The context menu for the shortcut contains the items in the context menu of the object itself plus the item D elete Short Cut Selecting D elete Short Cut will delete only the shortcut and not the object itself Selecting D elete will delete both the shortcut and the object itself To move a folder from one folder to another in the same O bject Browser window simply drag its icon to the other folder To copy instead of moving press CTRL while dragging the icon to the other folder To copy a folder from one folder to another in a different O bject Browser window simply drag its icon to the other folder To move instead of copying press ALT while dragging theicon to the other folder To delete a folder right click on the folder and press D elete Warning 154 The folder merely provides a link to objects If the database where these objects reside is detached then the objects will not appear in the folder EDITING OBJECTS AND DATA MANIPULATION EDITING OBJECTS AND DATA MANIPULATION Object
370. le 21 4 The Commands window toolbar Using a Command Line Create object oriented editable graphs The command line works in an interpretive fashion That is whatever you type in S PLUS will attempt to interpret and act on For example it can be used as a calculator gt 3 7 1 10 Note in the above example that spaces are redundant there can be an arbitrary amount of white space between the and the 7 The output from any S PLUS command is itself an S PLUS object in order that it can be used as input to a further command The 1 above simply means that the 561 CHAPTER 21 WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS Exiting the Commands window 562 following output starts at position 1 in the output object The key operation in S PLUS is an assignment which basically names the output in such a way that it can be used in the future It is extremely important to note that S PLUS stores a permanent copy on disk of any objects created in this way Assignment is carried out with the lt operation For example gt assigns 10 to x This is of course a trivial example Assignments can be made xX lt 347 to any S PLus data object such as data frames vectors matrices and lists A few characteristics of the S PLUS language are Clicking on the close window box closes the Commands window or to quit It is case sensitive the object X is not the same as object x
371. le panel simply select the columns and click on the time series plot button H owever in the graph created it very difficult to see variations in the unemployment and in the inflation rate because the axis scale is inappropriate Adding one or more right y axes can solve the problem but then the graph becomes cluttered with three possibly overlapping lines An alternative is to use the By Plot option Select the graph and click on the button in the middle of the last row of the 2D Plot palette to get separate panels with varying y ranges Each series will be plotted in separate panel The x axis range is held constant across panels but the y axis range is allowed to vary This is done by setting the Vary Axis Range property of the y axis You can also use multipanel graphs to display 3D plots rotated at various angles and sliced along axes Each panel shows a different view or section of the original surface For more information see section Displaying 3D Multipanel Graphs page 256 later in this chapter 237 CHAPTER 10 FORMATTING A GRAPH FORMATTING PANELS 238 Whenever a plot is placed on a multipanel graph a plot panel object is created for each panel T hey are contained by the plot in the O bject Browser and their properties can be changed by right clicking on their icon T hese panel objects have display attributes of their own that can be used to override the plot s display attributes For example you might want to set the
372. le windows of the same sheet or script open at the same time You can create a new window on your Data window for example by choosing New Window from the Windows menu The title bar of the original window will have 1 appended to it and the second window will have 2 appended to the window title D uplicate windows let you focus on different parts of a sheet at the same time Any changes made to the duplicate window will also be made to the original sheet window There is no limit to the number of times a window can be duplicated You have several options for viewing multiple windows To view the windows one on top of the other with only the title bar visible w From the Window menu choose Cascade To view the windows one on top of the other with the window visible w From the Window menu choose Tile H orizontal To view the windows side by side w From the Window menu choose T ile Vertical To close a window w From the File menu choose C lose or double click the C ontrol menu box in the upper left corner of the window You will be prompted to save any changes if you have not already done so To close all open windows w From the File menu choose Close All You will be prompted to save USING MENUS DIALOG BOXES AND TOOLBARS Using Main Menus Using Shortcut Right Click Menus each sheet or script if you have not already saved your changes S PLUS menus change depending on the type of window you are working on For e
373. leave this blank Options Deg of Freedom 377 CHAPTER 16 SUMMARIZING AND EXPLORING DATA 378 Results Enter a number for the degrees of freedom trace S where S is the implicit smoother matrix If both Degrees of Freedom and Smoothing Parameter spar are specified Smoothing Parameter spar is used unless it is 0 in which case D egrees of Freedom is used Smoothing Parameter lamda Enter a numeric value for the coefficient of the integrated second squared derivative penalty function If the value of spar is greater than 0 it is used as the smoothing parameter By default spar is 0 and D egrees of Freedom is used to control the smoothing If Degrees of Freedom is missing cross validation is used to automatically select spar Cross Valid Score Choose either generalized or ordinary cross validation score By default the generalized cross validation score is computed Use Unique x Values for Knots Check this to use the unique values of x as Knots By default a suitable fine grid of knots is chosen usually less in number than the number of unique values of x Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period
374. les N otes on Importing dBase Files N otes on Importing and Exporting Access Files N otes on Importing FASCII Formatted ASCII Files Importing OD BC Tables Exporting D ata Sets Exporting OD BC Files Chapter 6 Using The Object Browser Overview of the O bject Browser The Right Pane V ersus the Left Pane Filtering O bjects and D atabases T he Browser Page V ersus the O bject Browser Organization of O bjects The D efault O bject Browser T he Examples O bject Browser To Find S PLus O bjects Shortcut keys Customizing the O bject Browser Customizing the Right Pane D isplay Customizing O bject Browser Pages U sing Folders to O rganize W ork Editing O bjects and D ata M anipulation Object Creation M odifying D ata O bjects M oving and Copying O bjects Copying D ata from O ne D atabase to Another D atabase D eleting O bjects M odifying O bject Properties Selecting O bjects Chapter 7 Exchanging Objects with Other Applications Overview Embedding Objects from O ther Applications 123 126 126 127 127 127 128 128 128 130 135 138 141 141 142 143 144 144 146 146 146 147 148 149 151 153 155 155 155 156 156 156 157 157 159 159 161 vii CONTENTS Creating and Editing Embedded O bjects Importing Graphic mages Linking D ata from O ther Applications Embedding S PLus Graphics Within Other Applications U pdating Embedded Graphs Creating a PowerPoint Presentation Chapter 8 Creating a Graph S PLUS grap
375. les separated by s An intercept is automatically included by default For example Fuel Weight Disp fits a regression model with Fuel as the response and Weight and D isp as predictors For more information on formulas see the chapter on Building Formulas Create Formula Click this to open a formula builder dialog used to construct a formula specifying the desired model See the chapter Building Formulas for more information Family 457 CHAPTER 18 FITTING STATISTICAL MODELS Select a distribution family for the model Table 18 1 Certain combinations of Family and Link specify tandard statistical modds Save Model Object 458 Family Link Model gaussian identity ordinary least squares regression binomial logit logistic regression poisson log log linear regression Link Select the link function for the model The link function of the response is modeled as the sum of linear terms T he possible link functions depend on the family Variance For the quasi family a variance function can be selected Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are di
376. lick OK The new menu is visible in the O bject Browser Its name is the name of the external file without the extension SMN This example shows how to add to the context menu for objects of class dataframe displayed in the Object Browser The new item automatically computes summary statistics for the selected data frame To begin open an O bject Browser page and filter by Classlnfo and M enultem 1 Creating a Classinfo object for the Class data frame 1 Right click on Classi nfo default object and select C reate C lassI nfo in its context menu 2 Enter data frame in the Name field T his represents the name of the object class in which objects will have the context menu item specified below 3 Enter dfM enu in the Context M enu field This will be the name of the context menu 4 Click OK 591 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE 2 Creating the Context Menu 1 9 Right click on the Menultem default object and select Insert M enultem from its context menu Enter dfM enu in the Name field This corresponds to the Context M enu name given in to the C lassi nfo object above Enter M enu in the Type field Click OK Right click on dfM enu in the left pane and select Insert M enultem from the context menu Enter desc in the N ame field T his name is not important as long as it does not conflict with that of an existing object Select M enultem from the Type field Enter dataframe in the Docume
377. lick once on the Add Response button to change focus and once more to enter O zone as the response 78 IN DEPTH TOUR Select Radiation Temperature and Wind in the variable list Click on Add M ain Effect s to include these as predictors e Press OK to exit the Formula Builder dialog The formula you generated will be placed in the Formula field of the Linear Regression dialog Click OK By default a brief summary of the model will appear in a Report window and a four page Graph sheet will appear T he model will be saved as last Im More detailed results 1 Select Statistics R egression Linear 2 Use the history rollback button at the bottom center of the page to the right of Apply to select the previous dialog state T he previous values for data and formula will be filled in 3 On the Results page check the ANOVA Table check box This will provide an analysis of variance table for the linear model 4 Click OK The ANOVA table for the fit will appear in the Report window 5 Close the Graph sheet window when you are finished looking at the results 79 CHAPTER 3 TUTORIAL Pim Toim F Pebienuan a Pr r meh 7 ape Sed Prieta mi FA T ema E Fapparpa ag Pil bisbe ol Erena Erai Tip hnidy T papah Piare Ca Papo e Spm Fria Friii Phos T Fa Praba Ceri Figure 3 25 If on the Plot Page you check Residuals vs Fit and Repons vs Fit you g the multi page plot shown above Click th
378. link in your S PLUS Graph sheets By default data are linked to plots with automatic updating You can change this to manual updating in the Links dialog To edit links 1 From the Edit menu choose Links 2 In theLinks dialog select the link you want to edit 3 Choose Automatic or M anual linking If you choose M anual updating S PLUS only updates the link when you choose the U pdate N ow button from the Links dialog 166 EMBEDDING OBJECTS FROM OTHER APPLICATIONS Reconnecting or Changing Links If you rename or move the source file you need to reestablish the link In the Links dialog you need to rename the source file or specify the new location of the source file To re establish or change a link 1 From the Edit menu choose Links 2 Change the name of the linked source file or specify a different file name Choose OK 167 CHAPTER 7 EXCHANGING OBJECTS WITH OTHER APPLICATIONS EMBEDDING S PLus GRAPHICS WITHIN OTHER APPLICATIONS Because S PLUS supports object linking and embedding you can embed S PLUS Graph sheets within other applications such as PowerPoint and Word To embed an S PLUS Graph sheet 1 Load the client application e g Word and choose O bject from the client application s Insert menu Choose the Create New button and select S PLUS GraphSheet ChooseOK N ow you can create and activate the new Graph sheet 3 When you are finished editing the Graph sheet click o
379. lors 304 data specifications 303 fill direction 304 individual y values 304 multiple column data 303 plot options 304 stacked data 303 ARIMA model 528 arranging existing graphs 192 Ascending Sort button 106 ASCII files 126 ASCII specifying a format string 126 ASCII specifying column names 126 asymmetrical error bars 316 auto errors bar charts 307 error bar plots 317 auto means bar charts 307 error bar plots 317 autocorrelation 526 autocovariance 526 automatic span selection 373 automatically filled dialog fidds 342 bar base 306 bar charts 304 auto errors 307 auto means 307 bar base 306 bar width 306 colors 306 confidence level 307 data specifications standard 304 error bar calculations 306 error bar cap width 307 grouped offset 306 horizontal 304 multiple y data columns 305 stacked y data 305 605 INDEX bar charts 3D bar width 335 bar direction bar charts 306 error bar plots 316 bar grouping 306 bar offset bar charts 306 bar width 3D bar charts 335 bar charts 306 barley data frame 218 begin end vector plots 327 binomial mode 400 biplot 497 blocks deleting 105 inserting 103 borders changing for each item 303 style 302 Box Plots 307 box plots data specifications 307 grouped 308 break at missings 301 break at symbols 301 Browser Page 144 brush and spin 204 Brush and Spin window 602 buttons modifying 569 C cap width bar charts with error bars 307 error bar plots
380. lotsin 3D space and then rotate the results You can drag and drop a 2D plot button onto a 3D plane or you Plots in 3D can select Graph from the Insert menu to create a projected 2D plot Space Figure8 1 Multiple2D plotsin 3D space To project a 2D plot using drag and drop 1 Create anew Graph sheet 2 Click the 3D Plots button on the toolbar to display the plot palette 3 There are six 3D plane combinations near the center of the 3D plot palette Drag one of the 3D plane buttons off the plot palette and drop it onto the Graph sheet A 3D graph is drawn and the plane is added automatically to the graph The plane is automatically positioned at the minimum or maximum position depending on which plane button you choose You can drag and drop additional 3D planes as desired 201 CHAPTER 8 CREATING A GRAPH 4 Click the 2D Plots button on the toolbar to display the plot palette 5 Drag and drop a 2D plot button onto the desired 3D plane As you drag the plot over a 3D plane the plane becomes highlighted because it is an active drop target 6 The plot icon is now linked to the 3D plane You can double click the plot icon to specify your data columns or you can drag and drop data columns directly from your data W hen you have specified the data the 2D plot is drawn on the specified 3D plane To project a 2D plot using the menus 1 Choose Graph from the Insert menu 2 Choose 3D for the Graph Type then
381. lowess 373 CHAPTER 16 SUMMARIZING AND EXPLORING DATA KERNEL SMOOTHER This dialog estimates a probability density or performs scatterplot smoothing using Kernel estimates To perform kernel smoothing Choose Statistics gt Smoothing gt Kernel Smoother from the main menu The dialog shown below appears Kernel Smoother oP x r Data Points at Which to Compute Estimates gt Data Frame lethanal x Minimum s Variable Maximum l E Variable Number of Points 4 r Smoothing Options rail I Use Specified x Values for Estimation Kemel box le ATA Ez Bandwidth 0 5 Heels 3 Save s last ksmooth Plot Smooth Cancel Apply current Help Data Data Frame Select or enter the name of a data frame gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame x Variable Select the x variable y Variable Select the y variable Smoothing Kernel Options Select a smoothing kernel Kernel specifications include 374 CORRELATIONS AND COVARIANCES Points at Which to Compute Estimates Results box a rectangular box e triangle a box convolved with itself e parzen the parzen function a box convolved with a triangle normal agaussian density function T he default smoothing kernel is box Bandwidth Enter a numeric value for the kernel bandwidth smoothing parameter All ke
382. lready exists in database 1 and this data frame has the same LINEAR REGRESSION number of rows as the number of observations used in the model fit then the saved values are appended to this data frame This allows you to keep fitted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used gt Tip You may want to specify the same data frame as on the M odd page T his allows easy plotting of the fitted values or reiduals with the original data Plot Page Plots Residuals vs Fit Fitted Values Check this to save the fitted values from the model in the object specified in Save In Residuals Check this to save the residuals from the model in the object specified in Save In T hese are the ordinary residuals the response minus the fitted value Model Fanii Pini Prett Fios Opara F Amida va Ft F incude Bront DO Gage Albee Aorkiigie sa Fi T ehde Asgpiu F Responses ws Fi Hunia of Eiramaa Pins To ident Racial Homa 0g C Raika Fi Sorsa Purial Rasidial Fian P Fiod Petia Rsicels F Cook s Dame r r F CEJ eaj a i sf cement 1 Check this to display a plot of the residuals versus the fitted values Sart Abs Residuals vs Fit Check this to display a plot of the sq
383. lready exists in database 1 and this data frame has the same number of 447 CHAPTER 18 FITTING STATISTICAL MODELS 448 Plot page Plots rows as the number of observations used in the model fit then the saved values are appended to this data frame T his allows you to keep fitted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used lined Chocts Aneiysis of Wanante aE Modea Opts Fla Fior Plot Opbons P Amiduge ss Ft P inchide Emoni r B be Aandiiabs wa Fi T Fehde Aisa plat DV Responses wn Fi Hunia of Erama Ponts To kenty F Roridiss Homa Oa d M RatiduakFi Seread Purial Rasidial Fikir P Plot Paia Apriduss l Cook s Cite r r r Residuals vs Fit Check this to display a plot of the residuals versus the fitted values Sart Abs Residuals vs Fit Check this to display a plot of the square root of the absolute values of the residuals versus the fitted values This plot is useful for checking for the constant variance assumption of the model Response vs Fit Check this to display a plot of the response variable versus the fitted values Theline y x is also drawn on the graph Residuals Normal QQ Check this to display aN ormal quantile quantile plot of the residuals Residual F
384. lt less than gt greater than lt less than or equal gt greater than or equal amp and l or not Examples Examples of selection conditions given by where expressions are where sex 1 amp age lt 50 where income benefits famsize lt 4500 where incomel gt 20000 or income2 gt 20000 where incomel gt 20000 amp income2 gt 20000 where dept auto loan Note that strings used in case selection expressions need not be enclosed in quotes unless they contain embedded blanks Wildcards or are available to select subgroups of string variables For example where account 22 where id 3 FILTER PAGE The first statement will select any accounts that have 2s as the 5th and 6th characters in the string while the second statement will select strings of any length that begin with 3 The comma operator is used to list different values of the same variable name that will be used as selection criteria It allows you to bypass lengthy OR expressions when giving lists of conditional values for example where state CA WA OR AZ NV where caseid 22 30 4200 Missing Variables You can test to see that any variable is missing by comparing it to the special internal variable _ mi ssi ng For example where income missing amp age missing Sampling Functions T hree functions are available for sampling The first samp_rand prop allows for simple random sampling
385. lute values of the deviance residuals versus the fitted values This plot is useful for checking for the constant variance assumption of the model 469 CHAPTER 18 FITTING STATISTICAL MODELS 470 Options Partial Residual Plots Response vs Fit Check this to plot the response variable versus the fitted values The line y x is also drawn on the graph Residuals Normal QQ Check this to create a Normal quantile quantile plot of the pearson residuals Include Smooth Check this to display a smooth curve computed with loess smooth on the Residuals vs Fit Sqrt Abs Residuals vs Fit and Response vs Fit plots See the on line help for oess smooth for details Include Rugplot Check this to display a rugplot on the Residuals vs Fit Sqrt Abs Residuals vs Fit and Response vs Fit plots A rugplot is a sequence of vertical bars along the x axis that mark the observed x values Number of Extreme Points to Identify Enter the number of extreme points that will be identified on the Residuals vs Fit Sqrt Abs Residuals vs Fit Residuals Normal QQ and Cook s Distance plots The row names from the data frame specified on the model page will be used to identify the points Plot Partial Residuals Check this to display partial residual plots for all the terms in the modal Include Partial Fit Check this to display the partial fit for the term Include Rugplot Check this to display rugplots on the partial residual plots A rug
386. mbed data or a graphic from another application for example Excel into S PLUS Graph sheets Embedded objects become part of the S PLUS Graph sheet 159 CHAPTER 7 EXCHANGING OBJECTS WITH OTHER APPLICATIONS You should embed data or graphics when the embedded information is not likely to change the embedded information does not need to be used in more than one document or the source document is not available for updating if it is linked Note In order to use links or embedding the source application must support object linking and embedding OLE For example Excel 5 0 data can be embedded or linked in an S PLu s Graph sheet because Excel 5 0 supports OLE 160 EMBEDDING OBJECTS FROM OTHER APPLICATIONS EMBEDDING OBJECTS FROM OTHER APPLICATIONS Embedding Objects Using Cut Copy and Paste Embedding Objects Using Drag and Drop W hen you use cut and copy to move data S PLUS temporarily stores the data on the Clipboard You can paste the data into another location in the same document into another document created with S PLUS or into a document created with another application You can also paste data graphic objects or pictures created in other applications directly into S PLUS When you copy and paste a graphic object created in another application into S PLUS it is embedded as an object W hen you copy text from another application and paste it into S PLUS e if you are in a data
387. me you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame T his allows you to keep predicted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Predictions Check this to save the predictions in the data frame specified in Save In Standard Errors Check this to store the pointwise standard errors for the predictions in the object specified in Save In Related S PLus language functions loess loess control anova loess predict loess plot loess 439 CHAPTER 18 FITTING STATISTICAL MODELS STEPWISE LINEAR REGRESSION Sup wine Linen Pegeession ae ia Model Remi Dra Stepping pioni Daia Fimo femma e Model Page Data Data Frame Select a data frame T his dialog is used to fit linear models in a stepwise fashion to data in a data To perform stepwise linear regression Choose Statistics gt Regression gt Stepwise Linear from the main menu The dialog shown below appears ighi F Prinia Treoe of Aa Pir Sebo Ares eh
388. mle ij V Omit Rows with Missing Values Rotation varimax I Use Covariance List as Input r Save Model Object Lovarance List m Formula Variables Formula NOx Cancel Apply current Help Model Page Data Data Frame Select a data frame gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame 493 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS 494 Formula Model Subset Rows with Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the modal If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Use Covariance List as Input Check this to use
389. mple gui Set Output WindowPreference choice Report will change the default output to be the Report window and gui Set Out put Wi ndowPreference choice Default will change the default output back to the default T he default behavior for sending the text output is 1 Operations in the Commands window append to output in the Commands window 2 Operations in the Script window append to the output pane 3 Other operations including M enu dialog append to the Report window Error Warning By default most error warning messages go wherever the text output goes messages Some error warnings for operations on GUI objects go to the message window Trickling Input Some old S functions such asreadline inspect and so on need user input via keyboard We refer to this as trickling input and it is handled as follows 1 Operations in the Commands window enter at the Commands window 2 Other operations launch the dialog prompting for input 565 CHAPTER 21 WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS 566 CUSTOMIZING THE USER INTERFACE Overview Toolbars and Palettes 568 Creating Toolbars 568 M odifying Toolbars 572 Displaying Toolbars 574 M anipulating Toolbars 575 Saving and O pening Toolbars 575 Dialogs 576 Creating Dialogs 577 M odifying Dialogs 582 Displaying Dialogs 582 Example T he Contingency Table Dialog 582 Menus 586 Creating M enu Items 586 M odifying M enu Items 589 Displaying M enu Items
390. mple if you have already selected a block but want to add one more column hold down the sHirt key and press the RIGHT direction key To select the entire Data window w Click in theupper left hand corner of the D ata window to the left of the column headers You can limit the scope of some menu options by first selecting the appropriate columns in a D ata window Most menu options will take effect on the selected columns For example if you select a column then choose Edit then Clear the data in the selected columns are removed If you select columns and choose an action dialog for example Copy the selected columns will automatically be filled in for the source column field SELECTING CELLS COLUMNS AND ROWS Selecting Rows To select a single column w Click in the column number or row header to select the entire column or row All of the cals in the column or row are highlighted To select multiple columns or rows hold down the left mouse button and drag the mouse across the headers for the desired columns or rows To select all columns in the Data window w Click the upper left hand corner of the D ata window To select a range of contiguous columns w Drag the mouse across the desired column numbers To select a range of discontiguous columns w cTa_ click in the column header for each column You can select rows or blocks in a Data window before choosing menu options To select one row w Click the row number To
391. must have the same number of levels as the first classification variable Apply Continuity Correction Check this to apply a correction for continuity See the on line help for mcnemar test for an algebraic definition of the continuity correction Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten The saved object will have classht est This must be a valid S PLUS object name any combination of MCNEMAR S CHI SQUARE TEST alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names Print Results Check this to print the results in the designated output window Related S PLus language functions mcnemar test print htest menuMcnemar 407 CHAPTER 17 COMPARING SAMPLES MANTEL HAENSZEL CHI SQUARE TEST 408 Data This dialog performs a Mantd H aenszel chi square test on a three dimensional contingency table defined by three classification variables in a data frame See on line help for mantel haen test for more details To perform a Mantel Haenszel chi square test Choose Statistics gt Compare Samples Counts and Proportions gt Mantel Haenszel Chi Square from the main menu T he dialog shown below appears Manioliacnsrels Chi Gqaee Teal aE Date F App Contre
392. n For example if a symbol is selected the option will be Save Symbol Properties as D efault T he properties of the selected object are now saved as the default values T he next time you create a symbol it will use these new defaults To specify general settings select G eneral Settings from the O ptions menu Display in Grid Choose whether to have S PLUS automatically open a data window after importing data Respond to DDE Requests Choose whether S PLUS responds to D ynamic D ata Exchange queries See the chapter on DDE in the Programmer s Guide Old Format for DDE Requests Choose whether S PLUS uses text formatting style from earlier versions of S PLUS when responding to Dynamic D ata Exchange queries Enable Smart Cursor Choose this option to have the cursor always move in the direction of the last movement when the ENTER key is pressed while entering data in cells Enable Tool Tips Choose whether to have yellow prompt windows or tool tips displayed when you pause the mouse over a toolbar or palette button CHANGING DEFAULTS AND SETTINGS Startup Page Computations Page Color Toolbar Show the toolbars in color or black and white Large Buttons C hoose whether to show large or small toolbar buttons Dialog Field Tool Tips Choose whether to enable having yellow prompt windows or tool tips displayed when you pause the mouse over a dialog field Dialog Field Status Bar Choose whether to enable having short descr
393. n setup exe in the root directory of the CD ROM Use the default settings for installation It is a good idea to turn off other applications in particular virus checkers while installing S PLUS because of known problems with the installation software InstallShield If you are running a 16 bit operating system such as Windows 3 1 or Windows for Workgroups 3 11 you will need to have version 1 30 172 or higher of the Win32s subsystem on your machine before you can install S PLUS W in32s is included in the Win32s directory on the CD ROM and may be installed by running setup exe in the Win32s disk1 directory Be sure to install W in32s before installing S PLUS INSTALLATION Note Installing and running S PLUS under W in32s will require approximately 50M B of combined RAM and swap file space If you encounter the message S_apiSyncC onnect Failure several times during start up try increasing the swap file size to 40M B in the virtual memory settings accessed through the 386 Enhanced icon in the control pand Network T his version of S PLUS may not be installed on a network server If you want Installation to run S PLUS on a network server contact your sales representative for a network license 17 CHAPTER 1 WELCOME TO SPLUS System Requirements 18 Minimum platform configuration 486 IBM compatible PC Pentium recommended running at 66 MHz or more with 32M B of memory and math coprocessor H ar
394. n 155 O bject Creation 155 M odifying D ata O bjects 155 M oving and Copying O bjects 156 Copying D ata from O ne D atabase to Another D atabase 156 D eleting O bjects 156 M odifying O bject Properties 157 Selecting O bjects 157 The S PLUS environment is object oriented in that all items are distinct editable objects T his includes not only data objects and functions but also graph objects and interface objects such as menus dialogs toolbars and toolbar buttons The Object Browser is a imple and powerful interface through which to select view and edit objects created and used by S PLUS It 141 CHAPTER 6 USING THE OBJECT BROWSER operates similarly to W indows Explorer Table 6 1 The Object Browse toolbar z actErowzer Create Browser Expand Item Collapse Find Page Item S PLUS objects The Right Pane The Object Browser window is split into two panes that provide different views of the objects their sub components and attributes The left pane is a Versus the Left hierarchical display of objects For example an S PLUS list is a generalized Pane vector consisting of one or more components any of which could be another list This nesting of lists within lists can go on to any desired depth T he O bject Browser left pane allows the user to drill down to all underlying objects in a nested list A node representing an object can be expanded or collapsed to expose or hide the contents of an object and in
395. n Property object can only occur once in this list If pages or group boxes are specified it is not necessary to specify the Property objects that they comprise Property objects in the list will be displayed in two columns moving in order from top to bottom first in the left hand column and next in the right hand column Argument List Enter a comma separated list in the form 0 PropN amel 1 PropName2 Here PropNamel PropName2 are names of Property objects not including page and group objects and 1 refer in order to the arguments of the function indicated in Function Name The argument names may used in place of 1 2 Thefirst item 0 refers to the returned value of the function Use Argument List if the order of the Property objects in the dialog differs from the order of the corresponding arguments of the S PLUS function Prompt List Enter a comma separated list of labels for the Property objects DIALOGS in the dialog T hese will override the default labels T he syntax for this list is the same as that for Argument List Default Value List Enter a comma separated list of default values for the Property objects These will override the default values of the Property objects T he syntax for this list is the same as that for Argument List CallBack Function Enter the name of a function which will be executed on exit of any Property object in the dialog Help Command Enter command to be executed when t
396. n S PLUS graphs are stored in Graph sheets You can place one or more graphs on a Graph sheet and you can work with multiple Graph sheets A page on a Graph sheet represents a piece of paper on a desktop Graphical objects can be placed on the Graph sheet or anywhere on the desktop and are saved when the Graph sheet is saved to disk H owever only objects on the Graph sheet are printed The desktop area is useful for storing objects that may be needed in the future W hen you save an S PLUS Graph sheet all graphs and graphical objects you have placed in the Graph sheet are saved to a file extension SGR Graph specifications including plot type data references and any formatting specifications are also saved T he actual data however are typically not stored within the graph if you remove the data set the plot cannot be redrawn S PLUS is structured this way so you can create multiple graphs that reference the same data without saving a copy of the data with each graph See the section Changing Columns and D ata Sets page 188 for information on storing data with a graph Graph sheets may be exported to a variety of different image file formats including Windows Metafile WMF Windows Bitmap BM P and Tagged mage File Format T IFF T hese exported files can be imported into other applications You can save your Graph sheet in as many different file formats as you wish You must save your Graph sheet as an S PLUS graph file
397. n a Data window has a number T his number is displayed at the top of the column and shows the column s position in the D ata window Columns can also have names To add or edit a column name double click in the column s header Columns can be referred to by either their column names or column numbers It is usually wise to use column names because some operations can cause columns to be renumbered For example if you insert a column between columns 5 and 6 all columns to the right of column 5 are renumbered If you were using numbers to refer to these columns you would have to use the new numbers for subsequent operations SELECTING CELLS COLUMNS AND ROWS Column Lists in the Data Window Selecting Columns from a Menu List A column list isa list of column names or numbers used to specify a column or sequence of columns to be operated on You can refer to columns by their names or numbers or both H ere are sample column lists AGE AMOUNT 3 7 An entry can be a name a number two numbers separated by a colon a sequence of names separated by commas or the special key word ALL The sample column lists above refers to columns AGE and AMOUNT or the columns 3 through 7 Column names cannot be used to specify a sequence W hen prompted for a column list in a dialog field you do not have to type in the column names just click the arrow next to the fidd for a menu list of available column names Select the column you want pl
398. n an ppr PTET ai pr C r ne m Comia E ap O O Ema Joer Darah Ea Cig Hise Fe Gate r a F re O OOOO Figure5 4 TheExport D ata dialog D ata Specs page above O ptions bdow In the Export D ata dialog the D ata Specs page has the following options Data Specs page In the Export Data dialog the D ata Specs page has an upper section that allows you to navigate using the standard Explorer interface to a particular directory and file to be exported If the file already exists you will be warned and have an opportunity to overwrite it or cancel the export Exporting data You can export your S PLus data set asan ASCII file ASCII files can be read to ASCII files by most applications To export a data sheet as an ASCII file From the File menu choose Export D ata then To File Specify a file name for the exported data and the data set to export then choose OK 136 EXPORTING DATA SETS Options page In the Export D ata dialog the O ptions page has the following options Delimiter Choose a default delimiter to separate fields Specify a comma tab t a space or enter your own delimiter User defined delimiters must contain 8 or fewer characters If you leave the field empty the data is exported with fixed length lines see the Line Length field below Each field is formatted using column formatting information from the data set to determine the width and alignment and is blank filled Column N
399. n and choose Save as Default from the shortcut menu 115 CHAPTER 4 USING DATA WINDOWS 116 IMPORTING AND EXPORTING DATA Importing D ata Files 117 TheD ata Import Dialogs 119 Filter Page 123 Notes on Importing Files 126 N otes on Importing ASCII Delimited ASCII Files 126 N otes on Importing Excel Files 127 N otes on Importing Files with M ultiple Tables 127 N otes on Importing Lotus Files 127 N otes on Importing dBase Files 128 N otes on Importing and Exporting Access Files 128 N otes on Importing FASCII Formatted ASCII Files 128 Importing ODBC Tables 130 Exporting D ata Sets 135 Exporting OD BC Files 138 Importing One easy method of getting data into S PLUs for plotting and analysis is to Data Files import the data file S PLUS also allows you to export your data sets and graphs to many file formats for printing and for use in other applications Data Import In addition to ASCII and Formatted ASCII FASCII data file types you can Filters select from the following file types to import into S P Lus e Microsoft Excel versions 2 1 through Excel 97 XLS files Quattro Pro Worksheets W Q 1 Paradox D atabases D B 117 CHAPTER 5 IMPORTING AND EXPORTING DATA 118 Lotus Worksheets WKS WK1 WK3 W K4 and WRK dBase files D BF II II III IV files FoxPro files use same import filter as dBase files above Systat files double or single precision SYS files SigmaPlot files JN B
400. n expression that adds columns one and two together and then places the results in column three For more information on this option see the Programme s Guide When you want to edit a data cell you can double click on it or go to the cell and press ENTER to go into Edit mode or just start typing to overwrite the current data in the cell In Edit mode The RIGHT and Lert direction keys move the cursor character by character within the entry e The Backspace Key deletes the character to the left of the cursor e ThebeLete key deletes the character the cursor is on e Theesc key throws away all changes and exits Edit mode W hen you want to overwrite the contents of a cell type the new entry in the cell without entering Edit mode To erase the contents of a cell highlight the data in the cell with the mouse or by pressing ENTER and then press DELETE T he deleted cell is replaced with an N A which denotes a missing value To erase the contents of one or more columns highlight those columns and press D elete or select Clear from the Edit shortcut menu To edit existing data in a cell 1 Doubleclick the cell or select the cell and press ENTER 2 Use the direction keys to move among the data characters use BACKSPACE and DELETE to remove data characters and add any desired characters 97 CHAPTER 4 USING DATA WINDOWS 3 Press the ENTER UP Or DOWN key to accept the new data To undo an edit in a data cell w T
401. n option from the list box Function to do a linear or power Constant transformation Enter a number in the Constant box then click on the Do button to perform the transformation Example To create a formula for the linear regression dialog the user has to select a data frame from the D ata Frame drop down list box in the M odel page T he user can either type in the formula in the wide string box Formula or open a formula builder to do this For example you want to use the data frame fuel frame and run a linear regression The intention is to regress Fuel on Weight and M ileage 1 O pen the formula page After opening the Linear Regression dialog you choose the fuel frame from the D ata Frame drop down list box and click on the button Create Formula A formula page is open for building the formula Select a variable O n the formula page the variable names of the data frame fuel frame are listed in the list box Choose Variable s To set the response of the model select Fuel from the Choose Variable s list box Set the response Click on the button Response to set the chosen variable as the response Fuel is automatically placed in the wide string box Formula Select more than one variable To set the main and interaction effect of Weight and M ileage as predictors you need to press and hold the CTRL while clicking on Weight and Mileage Two variables are Selected Set the predictor Click on the button M ainHnteract t
402. n print the content of a Script window file 1 Select the Script window to print 2 From the File menu choose Print Script 3 Usethecommon Print dialog to specify your print options and click OK the actual Print dialog you see will depend on which platform you use and which default printer is currently in use Or 1 Click on the Print button on the standard toolbar 2 A dialog will show up to confirm which Script window you want to print 550 THE SCRIPT WINDOW Stopping a Script Interpreting Errors and Warnings Selecting text in a Script window Clearing Cutting Copying and Pasting text in a Script window Print Script 2 Printthe Script savetrees ssc 3 Click Yes to print using the default printer and default settings No to abort printing W hile running a script or a selected portion of a script you can usually stop it using the Esc key This will prevent the script from being evaluated any further If S PLUS encounters a problem with an expression or a command you enter in a script file it will display an error or warning in the output pane of the Script window T he warning or error message explains the problem and in some cases likely causes of the problem You can then move to this line and edit the script to correct the problem and re run the script Warnings are not considered as serious as errors Typically warnings will not stop script execution whereas errors will You ca
403. n select all text in the Script window by selecting Select All from the Edit menu or by pressing the CTRL A key You can move text within a Script window using the Cut Copy and Paste commands in the Edit menu or the Cut Copy and Paste buttons on the standard toolbar or the CTRL X CTRL C and CTRL V keys You can use these commands to move and copy text within the same Script window to another open Script window or between S PLUS and other applications Text that you cut or copy is placed on the Clipboard An item placed on the Clipboard will remain there until either the Cut or Copy command are chosen You can paste text from the Clipboard into a Script window as many times as you wish T he same techniques used to move and copy text are used within S PLUS to move and copy any item or character You can use the Clear command in the Edit menu or D elete key to delete text from the Script window without keeping a copy of the text on the clipboard 551 CHAPTER 21 WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS Using Undo in a Script Window Using Find and Replace 552 To move or copy texts in scripts 1 Select the text 2 Click the Cut or Copy button on the standard toolbar or choose Cut CTRL X or Copy CTRL C from the Edit menu This places the text on the Clipboard 3 Position the insertion point in a new location in the Script window Click the Paste button on the standard toolbar or choose Paste from
404. n the Font page you can specify standard font attributes for the polar axes labels ADDING MULTI LINE TEXT ADDING MULTI LINE TEXT In S PLUS you can add an unlimited amount of multi line text to your graph in the form of text comments main or subtitles axis titles and date amp time stamps To add multi line text 1 Choose Text from the Insert menu A text edit box will open O ther types of text for example Titles are also available from the Insert menu or 1 Open the Annotation palette and drag and drop the Comment icon see section Summary of the Annotation Palette page 277 onto the Graph sheet Click on the selected text to open the edit box or 1 Open the Annotation palette and click the Comment icon The cursor changes to the Comment Tool 2 On the Graph sheet click and drag the cursor Release the mouse button D efault text the size of the box is added to the graph 3 Click on the selected text to open the edit box Type the desired text Press ENTER to create a new line To end editing and save results click outside the edit box Alternatively you can press F10 Or press CTRL ENTER To exit without saving press the Esc key To edit existing text in place 1 Right click on the text and select Edit In place from the menu or 1 Click on the text to select it then click again 2 Make changesto the text in the edit box Press ENTER to create a new line To end editing and save results click
405. n the Licensed Works to MathSoft A Licensee may not translate decompile disassemble or reverse engineer the Software B Licensee agrees that because of the unique nature of the Software irreparable harm will be caused by a breach by the Licensee of its obligations hereunder that monetary damages will be inadequate to compensate for such harm and that MathSoft is entitled to injunctive relief to enforce this Agreement MathSoft s right to obtain injunctive relief shall not limit its right to seek further remedies C Licensee shall take all steps necessary to preserve and protect the propriety and confidential nature of MathSoft s software D The license and the warranties provided herein are extended to the original purchaser only and are not transferable E The licensee will not export or re export any part of the Licensed Works without the appropriate U nited States and or foreign government licenses
406. n the analysis To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use For example Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the moda If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Formula Enter a formula specifying the desired model The formula specifies which model is to be fitted In its simplest form a formula consists of the response variable a tilde and a list of predictor variables separated by s The difference between the formula for a generalized additive model and a generalized linear moda is that the additive model formula typically has smoothing functions applied to the predictors For example NOx C S E fits a generalized additive model with NOx as the response C as a linear predictor term and E asaspline smooth term For more information on formulas see the chapter Building
407. n the top left of the Object Browser frame is a drop down list with an edit control To filter objects using a string pattern enter the pattern in the edit control and select the M atch button to the left of the edit control Wildcards such as single character multiple character and ALL are acceptable The ALL wildcard denotes all objects are accepted and is the default T his is the string match setting of the O bject Browser shown in Figure 6 2 Each string pattern entered is saved up to amaximum of 20 entries in the drop down list and can be recovered T he pattern matching only applies to objects that are direct children of the root node in the left pane The different pages of an O bject Browser window can be made to display different classes of objects such as graph sheets data frames and functions T his is done by specifying an object filter for that page To create a new Browser Page first make sure the target O bject Browser is active Select the Create Browser Page toolbar button on the O bject Browser toolbar The Browser Page dialog will be displayed with the Filtering tab active A new Browser Page can also be created by the Insert BrowserPage menu item or by selecting the Create BrowserPage shortcut menu item right click on the white space of the left pane to bring this into view Alternately a Browser Page can be dragged from one Object Browser to another by dragging the tab of the source Browser Page and dropp
408. nce threshold for local scoring iterations Maximum Backfitting Iterations Enter the maximum number of backfitting iterations Backfitting Convergence Tolerance Enter the convergence threshold for backfitting iterations Print Iteration Trace Check this to print iteration details while the model is being fitted 467 CHAPTER 18 FITTING STATISTICAL MODELS 468 Results Page Printed Results Saved Results Geaerminod 4d deere Models Model Opiom Peate Fior Privted Pash Saved Renis C Stipu Sava h F Long Cigan V Frod vistas Vio rang Aendanis P Paaran Resadieds r Dima Asics l Response Pesjdtaig Short Output Check this to display a short summary of the modal fit to the designated output window T his includes the model call the degrees of freedom and the residual deviance Long Output Check this to display a detailed summary of the model fit to the designated output window T his includes the model call a five number summary of the deviance residuals the null and residual deviance values along with their degrees of freedom and the degrees of freedom for each term in the modd Save In Enter the name of an S PLus data frame in which fitted values and residuals areto be saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of rows as the
409. ne you can drag and drop one from the 3D Plot Palette or add one through the Insert menu A 2D plot can be attached to the projection plane To project a 3D plot onto a plane 1 2 Create a 3D surface plot From the Insert menu choose 3D Planes Choose an option from the submenu The 3D Plane dialog appears Click OK From the Plots 2D palette click and drag a2D Contour Plot button to a point immediately above the projection plane Release the mouse button to drop it A plot icon appears in the upper left corner of the Graph sheet D ouble click on the plot icon and fill in the appropriate fields on the Data to Plot page of the Contour Plot dialog Click OK 2D contours are drawn on the attached plane For a better view of the contour lines select the 3D workbox by clicking on the graph and rotate the graph using the green triangular knob 203 CHAPTER 8 CREATING A GRAPH BRUSH AND SPIN Quitting Brush and Spin Highlighting and Downlighting Points 204 Thebrush and spin window allows you to select points in one box in a scatter plot matrix and have them highlighted in all of the boxes In addition histograms can be drawn to reflect the highlighted and non highlighted points for each variable If you are examining three or more variables you can use the spin option to spin a3D point cloud projection To use brush and spin select Brush and Spin from the Graph menu A dialog appears with the following options
410. ng the class and inheritance of the data object catalyst 2 To enable the context menu for objects in the class design open the property dialog for the M enultem desc Enter data frame design in the Document Type field Click OK Return to the page showing data frames and right click on the object catalyst The context menu now contains Summary USING THE CLASSINFO OBJECT USING THE CLASSINFO OBJECT Overview Properties of the ClassInfo Object A ClassInfo object allows information to be specified about both user defined and interface objects It is smilar to the FunctionInfo object which allows information to be specified for functions primarily for the purpose of defining function dialogs T here are three main uses of the cl assI nfo object 1 Defining a context menu right click menu for objects 2 Defining the double click action for objects T hat is you can use it to specify what will happen when the user double clicks or right clicks on an object in the O bject Browser 3 It allows the dialog header and dialog prompts for interface objects to be overridden T he subcommand names of the properties are Name this must be the name of the associated class For instance to specify information for the I m class use this as the name This also becomes the name of this instance of thecl assInfo object ContextM enu this contains the name of the M enultem object that defines the contex
411. ning the following components x a vector of x data for line drawing y a vector of y data for line drawing Other Arguments For any of the smoothing types any of the optional arguments can be specified here For example if Friedman s supersmoother is used the underlying supsmu function is called If bass 5 is put into the Other Arguments field this will be passed down to the supsmu function when calculated Output Object Each underlying smoothing function returns an output object If you would like to save this object specify the name for the output object here You can add an extra dimension to line and scatter plots by varying the symbol sizes You can create a bubble plot by selecting data and choosing one of the bubble plot buttons on the plot palette Clicking one of these buttons uses a third column of data to specify the relative size of symbols on the Vary Symbols page of the Line Scatter Plot dialog For example if x is the gas mileage of a car and y is a measure of safety symbol size could vary depending on price The third column can be put in the z or w column on the Data To Plot page or specified on the Vary Symbols page You can add an extra dimension to line and scatter plots by varying the symbol colors You can create a color plot by selecting data and choosing one of the color plot buttons on the plot palette Aswith varying size clicking one of these buttons uses a third column of data to specify the relative c
412. nly the bottom triangle of the matrix plot drawn 329 CHAPTER 12 EDITING PLOT PROPERTIES Contour Plots 3D Data to Plot Gridded or Irregular Line amp Scatter Plots 3D Data to Plot Line Symbol 330 3D contour plots are identical to 2D contour plots except that the contours are drawn in 3D space at the level that they represent See section C ontour Level Plots 2D page 311 for plot property details Contour plots can use gridded and irregular data Gridded data can be specified in rectangular matrix form or stacked where the Z data is given in one long column and X Y and Z values are given in ascending order Irregular data is specified in a rectangular matrix form You will get better results if the data are distributed fairly uniformly in X and Y and do not contain sharp spikes or drops in Z See section Preparing Data for Graphing page 194 for details on specifying gridded and irregular data Stack 3D Contour Choose this option to have the contour plot stacked in 3D space If this option is not selected contours are drawn in 2D space For 3D line plots specify the names or column numbers of your X Y and Z columns all columns must be the same length For 3D line and scatter plots you can specify standard line attributes color style and weight and break options Pre Sort D ataline plots C hoose from N one X Y Z on X X Y Z on Y and X Y Z on Z to presort t
413. ns your plot palettes may look different than the ones shown in Figure 8 2 and Figure 8 3 5 A new Graph sheet is opened and the graph is drawn on the Graph 177 CHAPTER 8 CREATING A GRAPH 178 sheet Scatter Line Scatter w Line Line Isolated Points H igh D ensity Line w Text as Symbols Bubble Color Plot Bubble Color Loess Smoothing Spline Robust Linear Fit Polynomial Fit Exponential Fit QQ Normal Plot w Line BoxPlot Pie H istogram Density H istogram D ensity Bar Grouped Bar Stacked Bar Bar w Error Grouped Bar w Error Bar Origin Base Dot Plot H orizontal Bar Stacked H orizontal Bar 2D Time Series QQ Plot Contour Lower X Axis Left Y Axis No Conditioning Plots in Separate Panels Figure8 2 The2D Plot Palete H igh low Average Area Filled Contour Upper X Axis Right Y Axis 4 Panel Conditioning Sep Panels w Varying Y axis Vertical Error Bar Scatter Plot M atrix Levels Plot Upper X with Frame Right Y with Frame 9 Panel Conditioning Sep Panels w Varying X axis METHODS FOR CREATING A GRAPH Scatter Line Line with Scatter D rop Line Scatter Regression Regression with Symbols Coarse Surface D ata Grid Surface Spline Coarse Filled Surface DataGrid Filled Surface Filled Spline Surface 8 Color Draped Surface 16 Color Draped Surface 32 Color Draped Surface Bar Contour Filled Contour XY PlaneZ Min XZ PlaneY Min YZ Plane X Min XY Plane Z M ax XZ PlaneY M ax YZ Plane X M ax 2 Pane
414. nserting a N ew Column 66 Creating a2D Graph 67 Creating 3D Graphs 70 Using TrellisGraphics for M ultipanel Conditioning 74 Applying Statistics M odels 77 The Commands W indow 81 Creating PowerPoint Slides Automatically 83 Introduction S PLus is designed to work seamlessly with the software you already use You can import and export data from many sources including spreadsheets like Excel and Lotus analytical software such as SAS and SPSS and databases You can also access networked databases via O D BC O nce you have accessed your data you can analyze and explore it Let s walk through a sample session using some data to help you decide which new car you should buy ObjectBroweser Figure 3 1 TheFind button on the O bject Browser toolbar A note on using example data In this tutorial example data sets will be put in a folder in the default O bject Browser as they are needed T his is done using the Find option available on the O bject Browser toolbar see figure 3 1 Alternatively if you would like to 55 CHAPTER 3 TUTORIAL 56 browse through all of the example data sets you can load the example O bject Browser To do this close the default O bject Browser that appears at start up Choose File O pen from the Standard toolbar select Examples SBF and click Open To see a list of all the example data frames go to the left pane of the example Object Browser and click on the in front of data frame to expand t
415. nsitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last factanal is the most recent factor analysis modal fit Include Scores Check this to have the factor scores returned as a component of the fitted model Fattsi Annahar ol Model Optus Resia Pos Preda Medal Dpudrra Typa of Some epson E Siert ra valisa Maimun emim ho igqueenenss Tparangs ni Model Options Type of Score Select the type of factor score to compute the default is regression Starting Values Enter the name of a matrix of starting values for the maximum likelihood estimation procedure 495 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS Results Page Printed Results Loading Options 496 Maximum Iteration Enter a numeric value specifying the maximum number of iterations to perform for the maximum likelihood estimation procedure T he default is 20 iterations Uniqueness Tolerance Enter a positive number giving the tolerance for the change in uniqueness If no uniqueness changes by more than this value from one iteration to the next convergence is declared T he default is 0 0001 Fatto Anahi All E Model Opiom Penta Pos Predict Priksa Pascua ita L ni Opini FF Ghaap VC Component importance F bosingi Led ea a l a 1e Short Output Check this to print a summary of the model results in the designated output
416. nt Type field do not choose from the drop down box selections This corresponds to the object class which will have this context menu Select FUNCTION from the Action field 10 Enter the text Summary in the M enultem Text field T his text will appear in the context menu 11 M ove to the Command page of the dialog Tip A FunctionInfo object must exist for the function which is called by the context menu item O therwise the default dialog for that function will not appear 12 Enter menuDescribe in the Command field This is the function which is executed by the dialog which appears with Statistics D ata Summaries Summary Statistics There is a built in Functi onl nfo object by the same name 13 Show Dialog On Run This should be checked 14 T he M enultem object desc is now found alongside dfM enu in the 592 M enultem tree To move it underneath dfM enu hold down the Alt key and drag the desc icon onto the dfM enu icon To see the desc Menultem object in its new position click on the dfM enu icon in the left pane and look in the right pane MENUS 3 Displaying and Testing the Context Menu 1 Use File Open and select Examples SBF the examples O bject Browser 2 When data frame objects are visible in the right pane right click on the data frame named air Choose Summary which should appear under Properties in the context menu as shown in figure 22 17 Obpara Hroewurl
417. nt chooser dialog will let you select the font of your choice You can get help for any documented S PLUS function by right clicking on the function name and selecting Help The help viewer will open up at the given topic 557 CHAPTER 21 WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS TIME SAVING TIPS FOR USING SCRIPTS IN S PLus History Log 558 S PLUS provides several methods for writing scripts T he easiest is to open a new Script window and type in commands and execute them O ther ways to generate S PLUS scripts include using the History log and the menus or dragging objects into a Script window to record the commands that create or modify these objects T his section discusses how to view theH istory log how to generate a given plot using S PLUS commands and how to edit an S PLUS function s definition using a Script window You can drag and drop other object types from the Object Browser to a Script window Of particular interest are dragging toolbars menu items Classinfo and Function nfo objects respectively as examples of how to change which toolbar is showing modify the menu structure of S PLUS and create or modify dialogs in S PLUS respectively S PLUS keeps a continuous record or history of menu toolbar and dialog operations Visual edits such as changing cells in data windows or repositioning an object on a Graph sheet are also recorded as are commands issued in the Commands window The S PLUS programm
418. nt to have the height of the bars determined by the percent of data that falls within the interval The sum of the height of all of the bars is equal to 100 Lower and Upper Bounds Specify the lower bound minimum and upper bound maximum for the data you wish to use The default Auto automatic calculation uses the minimum and maximum values in the data provided Interval Width Specify the interval of the x axis for each bar of the histogram If the number of bars is specified the interval is computed automatically T he result will depend on whether Continuous or Integer has been selected For example suppose your data ranges from 1 to 10 and the interval width is specified as 1 With integer data a bar is drawn for each integer value from 1 to 10 10 bars are drawn W ith continuous data a bar is drawn for each interval of length 1 between 1 and 10 9 bars are drawn Number of Bars Specify the number of bars instead of specifying the interval of each bar The width of the interval is then determined by the number of bars and the range of the data T he higher the specified number of bars the narrower the interval width PLOT PROPERTIES Histogram Bars Density Line Line amp Scatter Plots 2D Data to Plot Line Window Specify the type of window used in the computations Number of Output Points Specify the number of equally spaced points at which to calculate the density Window Width Specify the width of the
419. ntain rows common to both data frames If just Data Frame 1 is checked then the result will contain common rows and those unique to the first data frame Similarly for the second check box If both boxes are checked then both common rows and unique rows will be included in the result By default the suffixes 1 and 2 are used for common variables not used as matching variables in the first and second data frames respectively To change these defaults enter new suffixesin D ata Frame 1 and Data Frame 2 Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to aname that starts with last For example last merge is the most recent merged data frame Show in Data Window Check this box to display the new data frame in a D ata Window Related programming language functions merge data frame data frame chind rbind match RANDOM SAMPLE GENERATION RANDOM SAMPLE GENERATION Sampling Options T his dialog generates random samples or permutes the observations held in a data frame or vector To perform random
420. nter anumber If a numeric variable is to be binned it will be binned into this many bins each of equal width Print Results Check here to display the table of counts in the designated output window Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names The default isl ast table The number of columns in the data frame will be one more than the number of variables selected in the Variables field The number of rows will be the number of unique combinations of values of the variables T he last column will be named count and will contain the count of the observations having the specified combination of variable values Related programming language functions table MERGE TWO DATA FRAMES MERGE TWO DATA FRAMES T his dialog combines data from two data frames into a single data frame To merge two data frames Choose Data gt Merge from the main menu T he dialog shown below appears Mange Tar Dats Fame AmE Crete Frome Inciude fiorai Aiea in Dist Fuse 1 J FE M Tuig Free 11 Dan Faa 2 P Cesta fa 2 Poe Heche Sut be He Musicheng Lorean Lot Mich bu T A Coan
421. number of observations used in the model fit then the saved values are appended to this data frame T his allows you to keep fitted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used GENERALIZED ADDITIVE MODELS Plot Page Plots Fitted Values Check this to save the fitted values from the model in the object specified in Save ln Working Residuals Check this to save the residuals from the final additive fit Pearson Residuals Check this to save the pearson residuals T hese are standardized residuals on the scale of the response Deviance Residuals Check this to save the deviance residuals The sum of squares of these add up to the deviance Response Residuals Check this to save the response residuals T hese are the response minus the fitted value Geaqalroed Adder Models CEET Opkone Foraurs Pige Predict Pios Pans Residual Plot F Amida sa Pit P Pio Partia Apsidueds r B be Apniiabs wa Fi r F Responses ws Fi F E Apridisi Homa OD F Opaons F pede Smooth l bieleda Fusgptot Mumba of Earme Pointe To ldeathy Residuals vs Fit Check this to plot the deviance residuals versus the fitted values Sqrt Abs Residuals vs Fit Check this to plot the square root of the abso
422. numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the modal If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Formula Enter a formula specifying the desired model Create Formula Click this to open a formula builder dialog used to construct a formula specifying the desired model See the chapter on Building Formulas for more information Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last manova is the most recent multivariate analysis of variance model fit 489 CHAPTER 19 USING MULTIVARIAT
423. o Statistics for further description of these methods Confidence Level Enter the joint confidence level desired T his value should be less than 1 and greater than 0 Bounds Select upper and lower for confidence intervals For one sided confidence bounds select either upper or lower Error Type Select the error rate type If family wise is selected the probability that all bounds hold is the level specified in Confidence Level If comparison wise is selected the probability that any one preselected bound holds is the level specified in the C onfidence Level Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names T he default is last multicomp The saved object is of class multicomp See the on line help for mul ti comp for more information on objects of this type Print Check this to display results in the designated output window T hese include estimates of the linear combinations their standard error and the confidence limits or bounds Plot Intervals Check this for a graphical representation of the intervals MULTIPLE COMPARISONS Metiple Comparisons
424. o condition Weight and M ileage 59 CHAPTER 3 TUTORIAL on athird variable Type 4 Minimize any windows you have open except the D ata window and Graph sheet you may need to Cascade the windows first using CTRL Shift C Press CTRL SHIFT V to vertically tile your graph and D ata windows side by side 5 Select the Type column by clicking on its header Then position the mouse somewhere within the data so that the pointer arrow appears not the down arrow of the column header and drag and drop it on the target area a rectangle marked by dashed lines that will appear at the top of your graph 6 Maximize the graph window T he data are divided into subsamples and conditioned by Type N ow you can see additional relationships Sporty cars normally assumed to be gas guzzlers actually have among the highest mileage along with Small cars e Compact and Medium cars often touted for higher mileage get gas mileage similar to Large cars Figure 3 6 TheAnnotations button on the Graph toolbar Identifying and Now we can identify which are the best and worst mileage cars by labeling Labeling Data the data points Points 1 Click on the Annotations toolbar button see figure 3 6 on the Graph toolbar use View Toolbars Graph to show the Graph toolbar if it isnot already in view On the Annotations palette click on the Label Point button see figure 3 7 2 Click on the data point you wish to label T he data point is
425. o generate Choose Data if you want to have this value equal the number of valid observations in your data Predicted Value Minimum and Maximum Specify the range of X values over which you want the curve evaluated or choose Data to use the minimum and maximum of the X data to determine the range If you specify minimum and maximum value s and the number of predicted values Increment must be set to Auto Increment Specify an increment value or Auto For example if you have specified 1 for the minimum 100 for the maximum and 1 for the increment then Number of Predicted Values must be set to Auto it will be equal to 100 T he regression is computed by forming and solving the normal equations All computations are done in extended 10 byte and double 8 byte precision A separate algorithm is used for polynomial regressions for computational efficiency On the By Confidence Bounds page you can specify up to eight pairs of confidence bounds T hey are defined automatically for your convenience By default each pair have their line styles set to N one To specify on one or more pairs of confidence bounds 1 In the lower portion of the dialog select one of the default confidence bounds specifications 2 Inthe Level field edit the confidence level if desired the value must be between 0 and 1 for example 0 95 0 99 3 Specify the line color style and weight for the confidence bound lines 4 Repeat steps 1 through 3 fo
426. o include Weight and Mileage as the main effects and interactions in the model You will see Fuel Weight M ileage in the wide string box Formula D efine a special term T here are some special terms available in the drop down list box Term Category For example if you want to set all second order interactions among Weight D isp and M ileage as predictors instead of main effects and interactions on Weight and Mileage select Weight Disp and M ileage as described in Step 4 N ow select the all cross terms from the Term Category and then click on the button Add The term is added to the formula You will see F uel Weight D isp M ileage 2 in the Formula box Remove a term To remove the intercept from the moda click on the check box Remove Intercept A 1 is added to the formula To 539 CHAPTER 20 BUILDING FORMULAS remove aterm for example Weight from the formula select Weight from the drop down list box Term and click on the Remove button Theterm Weight is removed from the formula Perform a transformation If you want to apply a log transformation on Fuel as the response you can select the variable Fuel from the Choose Variable s and click on the button Transformation The transformation dialog is opened Click on the button Log log x to display log Fuel in the New Variable s box Click on the button Add the new variable log Fuel is added to the Choose Variable s list box Click on the OK button to
427. of data points must be at least 2 but smaller than the number of input data points You can specify the desired font color size and styles for your contour labels These specifications are ignored for level plots You can also specify the following options Use Contour C olors Choose this option to have your contour labels use the contour line color For example if you have a red contour line the contour label will also be red When you change a contour color the contour label color will also change automatically Label Frequency Specify the frequency at which the levels are labeled For example a value of 3 indicates that every third level is labeled To place a label at each level enter 1 if you choose 0 no contours are labeled Occasionally S PLUS simply cannot find a good place on the contour to put a label T he success rate of the labeling can depend on the curvature of the data and on the number of grids specified Format Display Specify the display format type for the labels D ecimal Scientific Mixed or Auto If you choose Auto S PLUS chooses the optimal display format for the labels based on the data Precision Specify the number of digits to display after the decimal for the contour labels You can increase the precision to get more accurate labels for contour levels PLOT PROPERTIES Contour Fills By Contour Curve Fitting Regression Minimum and Maximum Level Specify the minimum and maximum values
428. of rows blank For example if you want labeling to begin at the second tick mark leave the first row in the column blank and enter your labels starting with the second row D ata Set Specify the name of the data set containing the label column Plane Specify whether to draw the tick labels on the XY XZ or YZ plane Skip First Label Skip Last Label Specify whether you want S PLUS to skip drawing the first tick label the last tick label neither tick label or both labels These options are useful if you have overlapping tick labels at axis intersections FORMATTING 3D PLANES FORMATTING 3D PLANES To select a 3D planes w Click anywhere on the grid plane To add a 3D plane Figure10 4 The3D plane combinationsfrom the 3D plot palete w There aresix 3D plane combinations at the bottom of the 3D plot palette Drag one of the 3D plane buttons off the 3D plot palette and drop it inside the graph area A 3D plane is drawn on the selected graph The plane is automatically positioned at the minimum or maximum position Or 1 select the graph 2 From the Insert menu choose 3D Planes 3 From the 3D Planes submenu choose XY Min XY Max XZ Min XZ Max YZ Min or YZ Max A 3D plane is added to the selected graph To create a 3D graph with a plane w Dragoneofthe3D plane buttons off the 3D plot palette and drop it onto a blank area of a Graph sheet A 3D graph is automatically created on the Graph sheet and a 3D plane is
429. of the contour levels T his information is used to set the range in the units of your Z data that the contours should represent If Auto is specified the minimum and maximum values of Z data are used for the range Contour labeling is improved if the number of contours divides evenly into the difference between the minimum and maximum levels Number of Levels Specify the number of contours Use Levels Column Choose whether to use a levels column for determining the contour levels Levels Column optional Specify the name of a column containing specific non uniform levels in the units of your Z data to be drawn The values should bein ascending order T he levels column specification is used in place of specifying the minimum maximum and number of contour levels if all are specified the levels column specification takes priority and is used to create the grid levels D ata Set Specify the name of the data set containing the levels column D raw Lines Choose whether to draw the contour lines If this option is not selected the line color style and weight specifications on the By Contour page for each contour will be ignored Fill Style Select None to not fill the contours with color Select Contour to have the fill areas defined by the contours Select Grids to create a level plot where each rectangle defined by the visual grid is filled with a color Contour will create a smoother plot while Grids typically draws faster If N
430. of the test statistics Save As ONE SAMPLE WILCOXON TEST Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last lm is the most recent linear regression model fit Print Results Check this to print the results of the W ilcoxon test in the designated output window Related S PLus language functions wilcox test t test 383 CHAPTER 17 COMPARING SAMPLES ONE SAMPLE KOLMOGOROV SMIRNOV TEST This dialog performs a one sample Kolmogorov Smirnov Goodness of Fit test of the empirical distribution against a hypothesized distribution To perform one sample Kolmogorov Smirnov test Choose Statistics gt Compare Samples gt One Sample gt Kolmogorov Smirnov GOF from the main menu T he dialog shown below appears One amplo KolmngooySmimoy Goodenss ol F i Tesi Cinta a Fen bale g Hypotheses Alem aie Hpo r aded Hypothesired Dest roma Parangiens ofthe Eoketod Diniibutioa m Data Data Frame Select a data frame
431. of the variables in the modal If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Formula Enter a formula specifying the desired model Create Formula Click this to open a formula builder dialog used to construct a formula specifying the desired model See the chapter on Building Formulas for more information Distribution Select the assumed distribution for the transformed response variable Link Select the transformation to be used on the response Fixed Parameters Enter a comma delimited list of fixed distribution parameters T his is most often just the scale for example scale 1 Type of Censoring Specify the type of censoring right left counting interval or interval2 Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he PARAMETRIC SURVIVAL only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last survreg is the most recent parametric survival model fit Paranentic mal Pilg Model Opiore Fimu Dpimo
432. ol is for read only Option List D elim Specify a character used as the delimiter for O ption List such as comma colon or semi colon Comma is the default Creating FunctionInfo objects Open the O bject Browser to a page with filtering set to FunctionI nfo 579 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE 580 Function Information 20 Function Prompt List ny Dialog Header Linear Regression DefaultValue List f 2 4 Status String zi CallBack Function backLm Property List Igui_Im_model Igui Help Command OoOo O Argument List 0 5PropRetum IV Write Arg Names Cancel Enpi f arn Figure 22 9 The dialog for a Function nfo object Right click on the default object labeled Function nfo in the left pane and choose Create Function nfo from the context menu T he property dialog shown in figure 22 9 appears Tl Display Help To modify the default settings that appear in this dialog in the future right click on the default object icon choose Properties from the context menu fill out the dialog with the desired default values and click O K Function Name Enter the name of the S PLus function which will execute when OK or Apply is clicked in the dialog This is also the name of the Functionl nfo object Dialog H eader Enter the text that will appear at the top of the dialog Property List Enter a comma separated list of Property objects to be displayed in the dialog A give
433. olors of symbols on the Vary Symbols page of the Line Scatter Plot dialog The colors are specified as a range or as a list of special colors if the third column is numeric If it is a factor colors will be used in the order specified in O ptions Graph Styles 323 CHAPTER 12 EDITING PLOT PROPERTIES Dot Plots 2D Step Plots 2D High Density Plots 2D Text as Symbols Plots 2D Dot Plots plot independent data against categorical dependent data using gridlines to mark the dependent levels This behavior is controlled by the Draw Gridline Property on the Line page of the plot dialog It is also necessary for the y axis State property for grids to be set on Auto A step plot uses alternating horizontal and vertical lines to connect data in steps You can create a step plot by selecting data and choosing one of the step plot buttons on the Plots 2D palette Clicking one of these buttons specifies the line connection type on the Line page of the plot property dialog as Vert First H oriz First H alf Vert First Half H oriz First Vert Only or H oriz Only For details see the description of the Line page of the plot property dialog High density plots draw a line between each data point and a specified axis You can create a high density plot by selecting data and choosing one of the high density plot buttons on the Plots 2D palette Clicking one of these buttons specifies the C onnect Type on the Line page of the plot
434. om defects in materials and workmanship MathSoft also warrants that the Software will be free from significant defects that prevent the Software from performing substantially in accordance with the accompanying documentation for a period of ninety 90 days from the date of purchase At MathSoft s option MathSoft will replace defective media and documentation fix significant defects in the Software without charge or refund the licensee fee paid to MathSoft by Licensee provided that the defective item is returned to MathSoft within ninety 90 days of the date of purchase Any replacement software will be warranted for the remainder of the original warranty period or thirty 30 days whichever is longer THESE REMEDIES ARE THE SOLE AND EXCLUSIVE REMEDIES AVAILABLE FOR BREACH OF EXPRESS AND IMPLIED WARRANTIES B THE FOREGOING WARRANTIES ARE IN LIEU OF ALL OTHER WARRANTIES EXPRESS OR IMPLIED INCLUDING BUT NOT LIMITED TO ANY WARRANT OF NON INFRINGEMENT MERCHANTABILITY OR LICENSE AGREEMENT AND LIMITED WARRANTY V Limitation on Liability FITNESS FOR A PARTICULAR PURPOSE IN ADDITION THE REMEDIES SET FORTH ABOVE WITH RESPECT TO A BREACH OF WARRANTY OR INFRINGEMENT SHALL BE THE EXCLUSIVE REMEDIES FOR ANY BREACH OF WARRANTY OR INFRINGEMENT HEREUNDER The sole purpose of such remedies is to provide Licensee with the repair or replacement of the purchased software or at MathSoft s option to refund the amount paid by Licensee hereunder
435. om many popular programs including SAS SPSS and Excel You can then transform manipulate and analyze your data using over 2 000 robust modern and classical functions In this example we will import data from the SAS file Exenvirn sd2 Importing a File To import Exenvirn sd2 1 From the File menu select Import D ata From File S PLUS lists the files in the current directory 2 Under Files of type choose SAS Files sd2 3 Navigate to your HOME directory this is your own named directory that contains your data and _ prefs sub directories Select Exenvirn sd2 4 Choose O pen to load the file into a window The Object S PLUS objects can be displayed in the O bject Browser To open the default Browser O bject Browser select the O bject Browser button on the Standard toolbar Figure 3 12 The Standard toolbar showing the postion of the O bject Browser button circled and the2D and 3D Plot palette buttons marked by the dli pse The default page shows the data frames lists matrices and vectors in the working directory any open Graph sheets and any open folders Select the data frame icon Note that the Exenvirn data frame that we imported is listed To view a data frame double click on its icon Pages can be added to the O bject Browser Each page can list a certain type of S PLUS object or a combination of objects Selecting tabs at the bottom of the window will switch the page viewed 65 CHAPTER 3 TUTORIAL
436. omit from the analysis any rows in the data frame that contain missing values for any of the variables in the modal If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Formula Enter a formula specifying the desired model The formula specifies which regression model is to be fit In its simplest form a formula consists of the response variable a tilde and a list of predictor variables separated by s An intercept is automatically included by default For example Fuel Weight Disp fits a regression model with Fuel as the response and Weight and Disp as predictors For more information on formulas see the chapter on Building Formulas Create Formula Click this to open a formula builder dialog used to construct a formula specifying the desired model See the chapter Building Formulas for more information Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names T he default is last Im T he saved object will have class I m Seethe on line help for I m ob
437. omplete the message information in your mail system and then send the message For more information refer to the documentation for your specific mail application Troubleshooting The Send command does not appear on the File menu Exporting Graph Sheets to Different File Formats Make sure you have installed M icrosoft Exchange on your system or any mail system compatible with the Messaging Application Programming Interface M API or Lotus cc M ail or any mail system compatible with Vendor Independent M essaging VIM e If you have installed Microsoft Exchange or a compatible mail system on Windows 3 1 make sure the WIN INI file contains the following lines or add them MAI L MAPI 1 If you have installed Lotus cc M ail or a compatible mail system make sure MAPIVI32 DLL and MAPIVITK DLL are in the Windows System folder You can export your Graph sheets to a variety of file formats for use in other applications See the chapter Exchanging Objects with Other Applications for more information on S PLUS s OLE capabilities To export a Graph sheet to a different file format 1 Make sure the Graph sheet you want to export is the current Graph 339 CHAPTER 13 EXPORTING GRAPHS PRINTING AND SENDING MAIL Sheet its title bar is highlighted 2 From the File menu choose Export Graph Choose the desired file type in the Save as Type box and type a file name in the File Name box 3 Choose Save to save the
438. on and ARIM A modals Refer to chapter 14 Using the Statistics Menus and Dialogs for an explanation of features common to most dialogs 487 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS MULTIVARIATE ANALYSIS OF VARIANCE This dialog performs multivariate analysis of variance the extension of analysis of variance techniques to multiple responses To perform multivariate analysis of variance Choose Statistics gt Multivariate gt MANOVA from the main menu The dialog shown below appears kemena Analysis ol Vanam Sl Ei Morel Opina Pee Dena Wigh E Subest Aran w Seen Mocks Obert E Oman Roses wiht Messing akm Shei Ai TETT Fomuls Faika Cane Forres Lick cancel priv i ferent 1 Model Page Data Data Frame Select a data frame gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame Weights Enter the column that specifies weights to be applied to all observations used in the analysis To weight all rows equally leave this blank Subset Rows with Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expression 488 MULTIVARIATE ANALYSIS OF VARIANCE Formula Save Model Object must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the
439. onents of sets of observations of multiple variables See also the chapter Principal Component Analysis in the Guide to Statistics page 463 To perform principal components analysis Choose Statistics gt Multivariate gt Principal Components Analysis from the main menu T he dialog shown below appears Principal Components Analysis Ioi x Model l Results Plot Predict Data m Model Data Frame ethanol Scaling Covariance Matr Subset Rows with Correlation Matri M Omit Rows with Missing Values Save Model Object I Use Covariance List as Input Save As last princomp Covariance List l M Include Scores Formula Variables All Variables NO Formula NOx Cance Apply current Help Data Data Frame Select a data frame gt Tip You can typeinto the D ata Frame edit box any expression which evaluates to a data frame Subset Rows with Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE 500 PRINCIPAL COMPONENTS ANALYSIS Formula Model Save Model values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only t
440. onvert To button to have S PLUS convert the embedded object to the format specified for the Object Type or click the Activate As button to have S PLUS open all embedded objects in the proper application for the format you specify for the object type 4 ChooseOK to convert the object If an embedded object was created in an older version of an application currently installed on your system you will need to convert it to match the current version of the application Some applications will convert existing files to the new format during installation of a newer version You may be able to run the application s Setup program to convert existing files You can convert embedded objects to a simple picture graphic Converting embedded objects to pictures sometimes reduces the size of your Graph sheet H owever you will not be able to edit the picture after converting it To convert an embedded object to a picture see the instructions in the previous section Converting to a Different Format and choose Picture for the O bject Type You can import graphics created in a variety of other applications into an S PLUS Graph sheet T hese graphics are embedded as pictures in the Graph sheet T his is different from an embedded object T hese pictures are smaller in size They cannot be converted back to embedded objects for further editing and they cannot be edited from within S PLUS EMBEDDING OBJECTS FROM OTHER APPLICATIONS Fill Border Position
441. ools Shapes and Symbols Drawing Objects You can use the mouse as a drawing tool to add drawing objects to your graphs or you can drag annotation objects from the palette onto your graph Use the Selection tool to deselect a drawing object and return to the selection Use the Comment tool anywhere on your graph You can also use the Text option on the Insert menu to add text to your graph For a more detailed 277 CHAPTER 10 FORMATTING A GRAPH 278 discussion of adding and editing text see the section Adding Multi line Text page 261 Date Stamp Use the D ate Stamp tool from the Annotation palette to place a default date stamp on your graph The date stamp contains default text and the current date and time Doubleclick the date stamp to edit its text or font specifications For more information on date stamps see section Adding a D ate and Time Stamp page 269 Arrows Use the Arrow tool from the Annotation palette to draw arrows T he starting point represents the tail of the arrow The point where you release the mouse button is where the head of the arrow is placed Lines Use the Line tool to draw straight lines Arcs Use the Arc drawing object to draw all or part of a circle Arcs will draw proportionally to an invisible center point therefore the closer you move in toward the center the smaller the arc will be Conversely the farther you move out the larger the arc Radial Lines Use the Radial Line
442. oose History then choose Clear from the submenu 559 CHAPTER 21 WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS Dragging Graph Objects into a Script Window Another way to create scripts is to drag objects from a graph like plots or extra symbols into a Script window If you want to know which S PLUS commands are used to create or modify a particular plot you can drag it into the Script window and the S PLUS command to create it will be written there automatically You can then run the generated script to create or modify the plot This is an alternative to using menu options and dialogs to create or modify To drag graph objects into a Script window 1 Open a previously saved Graph sheet or create a new Graph sheet containing the objects for which you want commands 2 Open a previously saved script or create a new Script window by choosing N ew from the File menu 3 You may wish to vertically tile the Script window and the Graph sheet window T his makes it easier to select and drag objects between the Graph sheet window and the script To do this choose Tile Vertical from the Window menu 4 Select the objects from the graph and drag them into the upper pane the program pane of the Script window Dragging the mouse As you drag the mouse the cursor changes into a drop cursor When the mouse is inside the program pane you will see a gray vertical marker line at the left most edge of the line you a
443. open Enter the data for the new object 348 TABULATE TABULATE This dialog creates a tabular summary of data from a data frame Selected columns of the data frame are identified as variables and the count of each combination of variable values is returned Numerical data can be optionally binned before the counting occurs The table of the counts can be printed and also returned in a data frame that is suitable for multi panel conditioning plots To obtain statistics and other summary information choose Statistics gt Data Summaries gt Crosstabulations from the main menu To tabulate data Choose Data Tabulate from the main menu The dialog shown below appears Dom Ophea ra Dais prame Madni Unique Huriet Walis i vaiablag z e Numier of Bins foe Hemana walu f Eele F Print Resets Sree An EEr Data Data Frame Select the data frame from which to select variables Variables Select the variables to be tabulated T he counts of each combination of values among the selected variables will be calculated gt Tip Leave the D ata Frame fied blank and jug typein the names of vectors in the Variables field for data thatisnotin a data frame Options Maximum Unique Numeric Values 349 CHAPTER 15 CREATING AND MANIPULATING DATA 350 Results Enter a number Numeric variables having more distinct values than this number will be binned Number of Bins for Numeric Values E
444. or Tick Labels labels for minor ticks T he options on all three pages are the same Pages Label Type Choose from the options listed in the table below to specify the label type The currency type and the language for month and day series options are specified in Regional Settings in the Control Panel Table 10 1 Axis laba types Label Type Definition None No tick labels drawn Column Choose to usea column of label names Auto S PLUS chooses the optimal display format for the labels based on the data Decimal Choose to specify your labels in decimal format Scientific Upper Choose to specify your labels in exponential form with a capital E for example 10E 007 Scientific Lower Choose to specify your labels in exponential form with a lowercase e for example 10e 007 M ixed Choose to use a mix of decimal and scientific formats for data with a large data range for example data ranging from 0 1 decimal to 10e 007 scientific Percent Choose to specify your labels as percentages for example 100 0 M in Precision Specify that all data has the same number of decimal places as the least precise value Base E xponent Choose to show the base and exponent for example 10 3 Exponent Only Choose to show the exponent only for example 3 for 1073 Currency Specify that the labels be in units of currency 245 CHAPTER 10 FORMATTING A GRAPH Table 10 1 Axis laba types Financial M onth Series
445. or the constant variance assumption of the model Response vs Fit Check this to display a plot of the response variable versus the fitted values Theline y x is also drawn on the graph Residuals Normal QQ Check this to display aN ormal quantile quantile plot of the residuals Residual Fit Spread Check this to display a residual fit spread plot T his is a visual analog of the multiple R squared statistic It compares the spread of the fitted values to the spread of the residuals RANDOM EFFECTS ANALYSIS OF VARIANCE Options Partial Residual Plots Cond Plots of Fitted vs Predictors Check this to display the conditional plots of fitted values versus predictors Include Smooth Check this to display a smooth curve computed with I oess smooth on the Residuals vs Fit Sqrt Abs Residuals vs Fit and Response vs Fit plots See the on line help for oess smooth for details Include Rugplot Check this to display a rugplot on the Residuals vs Fit Sqrt Abs Residuals vs Fit and Response vs Fit plots A rugplot is a sequence of vertical bars along the x axis that mark the observed x values Number of Extreme Points to Identify Enter the number of extreme points that will be identified on the Residuals vs Fit Sqrt Abs Residuals vs Fit Residuals Normal QQ and Cook s Distance plots The row names from the data frame specified on the model page will be used to identify the points Plot Partial Residuals Check this to dis
446. ormula for the model If omitted the original data are used for computing predictions Save Save lin Enter the name of an S PLUS data frame in which predictions and standard errors are to be saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame This allows you to keep predicted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Predictions Check this to save the predictions in the data frame specified in Save In Standard Errors Check this to store the pointwise standard errors for the predictions in the object specified in Save In Options Prediction Type 463 CHAPTER 18 FITTING STATISTICAL MODELS Select the type of prediction to be saved e link predictions are on the additive predictor link scale response predictions are on the response scale using the inverse link function terms a matrix of predictions is produced one for each term in the modal S PLus language functions related to Generali
447. osition the mouse within the selected region not in the column header until the cursor changes to an arrow Pressing the left mouse button drag the data and move it over a plot icon When the plot icon changes color release the mouse button to drop the data and generate the plot T he new plot is drawn on the graph using the dropped data If the data is not appropriate for the plot type or if there is insufficient data to draw the plot for example you drag two columns of data onto a contour plot icon an error message appears To add data to a plot icon using the plot property dialog 1 Doubleclick the plot icon to display the plot dialog or right click on the plot icon and select data to plot from the shortcut menu 2 In the Data to Plot page specify the data columns to be plotted 3 Make any other desired changes in the dialog and choos OK T he new plot is drawn on the graph using the specified data To replace data in an existing plot using drag and drop 1 Choose Tile Vertical from the Window menu so you can see your data and Graph sheet simultaneously 2 Select the data columns Use ctat click to select discontiguous columns 187 CHAPTER 8 CREATING A GRAPH Changing Columns and Data Sets 188 3 Position the mouse in the selected columns until the cursor changes into an arrow Pressing the left mouse button drag the data from the D ata window or O bject Browser and move it over a plot W hen the plot ch
448. oup in a grouped bar chart each slice in a pie chart each area in an area chart and each contour level in a contour plot An Auto can be placed in the text field by entering an For information on specifying special characters in the text field see section Adding Special Characters and Formatting Text page 265 earlier in this chapter Hide C hoose whether to display the legend item Override Auto Legend Item Specs Choose this option to specify custom line and symbol attributes for the legend item When you click on this check box the line and symbol attribute fields become enabled You can specify standard line attributes style color and weight and symbol attributes style color and weight for the sample If you do not have this option selected the legend item will be updated automatically when changes are made to the associated plot s properties To prevent the legend item from being deleted when the associated plot is deleted explicitly specify the number of legend items in the Legend dialog Override Auto Legend Item Specs C hoose this option to specify custom fill and border attributes for the legend item W hen you click on this check box the fill and border attribute fields become enabled You can specify standard fill attributes fill color fill pattern and pattern color and standard line attributes style color and weight for the pattern sample If you do not have this option selected the legend item will be
449. outside the edit box Alternatively you can press F10 Or press CTRL ENTER To exit without saving press the esc key To move text 1 Select the text Green selection knobs appear on the outline of the 261 CHAPTER 10 FORMATTING A GRAPH text box 2 Click indide the selected region and drag the box to anew location You can also move the text by changing the X and Y position on the Posi tion page of the property dialog available from the shortcut menu To resize text 1 Select the text Green selection knobs appear on the outline around the text box 2 Drag one of the square green knobs to increase or decrease the size of the box The proportions of the text ratio of height to width remain constant To rotate text 1 Select the text Green selection knobs appear on the outline of the text box A round green knob appears at the bottom of the selected text 2 Drag the rotation knob up or down The position of the lower left hand corner of the text box remains fixed You can also rotate the text by specifying a rotation angle on the Position page of the property dialog available from the shortcut menu To format text in place 1 Open the text edit box and select the text 2 Choose a Graph Sheet Toolbar option Font type and size listboxes Bold Italic Underline Superscript and Subscript buttons to change the format of the text You can change the font and point size and choose whether to bold italici
450. own number of conditioning variables list on the Standard toolbar 2 Select the x and y columns you wish to plot followed by the conditioning variable s Click on the plot type on the plot palette To create Trellis graphs using drag and drop 1 Create a graph 2 Drag acolumn of conditioning data onto the graph and drop it in the highlighted rectangle at the top of the graph the highlighted rectangle is shown in the figure above 3 If the conditioning data are continuous you can use the 209 CHAPTER 9 WORKING WITH TRELLIS GRAPHICS conditioning buttons to change the number of panels Figure 9 3 Dragging and dropping conditioning data To create Trellis graphs using Shift click 1 Create a graph and select the graph area 2 Select the conditioning data column s in the Data window or O bject Browser 3 sHiFt click one of the conditioning buttons on the plot palette Plots on Trellis Plots on Trellis graphs behave very much the way they do on standard graphs Graphs You can 1 Doubleclick or right click to change the data specifications or any other attributes W hen plot specifications change all of the panels are modified 2 Change the plot type by selecting the plots and clicking on the plot palette 3 Drag new data onto them 4 Add additional plots By default each plot uses the same conditioning variables specified in the multipanel page of the graph to determine which rows of the d
451. p e greater true parameter is greater than p Save As EXACT BINOMIAL TEST Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten The saved object will have classht est This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names Print Results Check this to print the results in the designated output window Related S PLus language functions binom test print htest menuBinom 401 CHAPTER 17 COMPARING SAMPLES PROPORTIONS TEST This dialog compares sample proportions against hypothesized values using Pearson s chi square test statistic See the on line help for prop test for details on the assumptions of this test D ata can be but need not bein a data frame To perform a test of proportions Choose Statistics gt Compare Samples gt Counts and Proportions gt Proportions Parameters from the main menu T he dialog shown below appears Data Proportions Tosi Den Dzis Frene o deceives al Trals minie E Hepvembis Data Frame Ahama Hypoth fess nce Gontdence Lem ps F pp raes Conni Consp Focus p Eirg Prev Feces 1 Select a data frame although this is not required see
452. p Select this to open a help page on toolbar buttons Toolbars can be modified using their property dialogs or their context menus Using the Property Dialog Right click on a Toolbar object and select Properties from the context menu The Toolbar property dialog appears as in figure 22 3 Toolbar Mhjact Siandand 1 al E a ee FF Doei Touber Laym Doira Docsi To TOF F Iopa z Tooter Top fas O Lang Bunions T nis Lat Sariy C H Burton Pires l Figure22 3 The Toolbar Property dialog 572 Document Type Depending on the type of document window Graph Sheet Commands window etc which has the focus a toolbar may or may TOOLBARS AND PALETTES not be visible Select the document types for which the toolbar should be visible The choice All Documents causes the toolbar to be always visible The choice N o Documents ensures that the toolbar will be visible when no document window has the focus for example when no window is open ColorButtons Check this to display button images in color ToolTips Check this to enable tool tips for the toolbar LargeButtons Check this to display large sized buttons Hide Check this to hide the toolbar This is also available through the Toolbar object context menu D eletable Check this to allow permanent deletion of the toolbar Docked To Select the side of the S PLUS window to which the toolbar will be docked or select NONE to float the toolbar as a palette Toolbar Top
453. perate on multiple windows making it easy to edit data run scripts and create graphs S PLUS windows have the same elements as most W indows based software The Control menu box is always in the upper left corner of the window Click once on the C ontrol menu box for a list of commands that control the size shape and other attributes of the window Click twice on the Control menu box to close the window The title bar displays the name of the window If more than one window is open the title bar of the current or active window is a different color or intensity than other title bars The minimize button is represented by a horizontal bar and when clicked reduces the window to an icon The maximize button is represented by a square and when clicked enlarges an application window to fill the entire desktop or will enlarge a sheet window to fill the entire application window 31 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE 32 SPL Dijar Groene Ble Gdn view aset Format Cota grisi Graph dpten Wade Help Djela 2 2 ele aff 19 OFF whl ale tof we sal e j Al e T Match i r Corimir al date iura Pam H i Cs es f Gimhae T eS emch ah Sok bey lban HH Figure2 2 Theopening main window of S PLUS brings up the default O bject Browser window N otice that both windows havea control menu box top left and minimize maximize and close buttons top right The O bject Browser window can be sized
454. ph O ptions dialog will appear Default to Draft Mode Choose whether to have your graphs displayed onscreen in draft mode You can also toggle this option on and off in the View menu Using draft mode will speed up redraw time dramatically D raft mode only affects screen resolution printed output will always be publication quality Condition Mode On Conditioning mode affects how selected data is used in creating a plot with a plot button When conditioning mode is on the last column s selected are used as conditioning variables for a multipanel graph See the chapter on Trellis Graphics The number of columns used for conditioning is defined by the of Conditioning Variables below Conditioning mode can also be turned on and off by using the conditioning mode button on the main toolbar of Conditioning Variables Specify the number of columns to use as conditioning columns when conditioning mode T hisnumber can also be set using the drop down list of numbers on the main toolbar Snap to Grid Per Inch Specify the number of invisible gridlines to use for the Snap to Grid option The default is 12 gridlines per inch or 5 gridlines per centimeter When Snap to Grid is enabled objects will snap to the closest intersection of the invisible horizontal and vertical gridlines Create New Graph Sheet C heck this box to have a new Graph sheet created when plots are created using a statistics dialog Resize Fonts with Graph C hoose whether ti
455. ph sheet to the clipboard 5 Select the empty target Graph sheet window 6 From the Edit menu choose Paste The contents of the copied Graph sheet will be pasted into the target Graph sheet window 7 Repeat steps 4 through 7 for each Graph sheet you wish to combine with the target Graph sheet Each time you add a graph existing graphs will be automatically arranged to accommodate it To add an additional page to your Graph sheet right click on the page tab bar at the bottom of the Graph sheet and select Add Page If you have already created some graphs you can use S PLUS s Arrange Graphs feature to position them on the Graph sheet When you choose a layout style S PLUS positions and sizes your graphs on the Graph sheet automatically To arrange existing graphs 1 From the Format menu choose Arrange Graphs 2 From the Arrange Graphs submenu choose the desired layout style D efault Size Position One Across Two Across T hree Across Four Across or O verlaid 3 This determines the number of graphs to place horizontally before wrapping graphs to the next row To specify the order in which the graphs are laid out you can select them in the desired order before using Arrange Graphs Use sHirt click to select multiple graphs and then choose a graph layout Be sure to select the entire graph area when selecting graphs PLACING MULTIPLE GRAPHS ON A GRAPH SHEET You can use Arrange Graphs as many times as you wis
456. phics It essentially searches for the three points nearest to each grid point and then finds where the plane through these points PLOT PROPERTIES 3D Bars Bar Charts 3D Spline Plots 3D intersects the vertical line that goes through the grid point that is the line perpendicular to the x y plane at the grid point This intersection is used as the estimate of the height of the surface at this point Weighted Average Used only for the internal algorithm S PLUS will calculated a weighted average of all points that lie within each grid cell before interpolating the grid height If your irregular data has many irregularities using the weighted average method may smooth them out within each grid cell As a result the weighted average method slows down plotting but produces a smoother contour If the average number of data points within each cell is less than three there is not likely to be any advantage to using weighted averaging Extrapolate U sed only for the bivariate algorithm If checked data points will be extrapolated outside of the convex hull determined by the data points filling the entire computed grid If not checked only the area of the plot corresponding to the convex hull of the original data will be drawn The internal algorithm always extrapolates Pts Partial Der Used only for the bivariate algorithm Percentage of additional points to be used in computing partial derivatives at each data point If set to
457. play partial residual plots for all the terms in the modal Include Partial Fit Check this to display the partial fit for the term Include Rugplot Check this to display rugplots on the partial residual plots Common Y axis Scale Check this to give all the partial residual plots the same vertical units T his is essential for comparing the importance of fitted terms in additive models S PLus language functions related to Analysis of Variance aov summary aov plot Im coef dummy coef raov Other related S PLUS functions Im manova mul ticomp 455 CHAPTER 18 FITTING STATISTICAL MODELS GENERALIZED LINEAR MODELS This dialog fits generalized linear models It calls the gl m function and its print summary plot and predict methods The class of generalized linear models includes many standard statistical models including ordinary linear regression logistic regression and log linear models T he specification of the family and link function determine which type of moda is fit To fit a generalized linear model Choose Statistics gt Generalized Linear Models from the main menu The dialog shown below appears Geeemhred Lincas Models Ei madai Optoma Remta Fia Fredis Osa Mois Daia Frama Family fasu ELI Weperg hts Link Subesi Aras with F One Pies at mg akne Saree hadai Obed Sata fim gm Famula Formula Crete Frmiia Model Page Data Data Frame Select a data frame
458. plot is a sequence of vertical bars along the x axis that mark the observed x values Common Y axis Scale Check this to give all the partial residual plots the same vertical units T his is essential for comparing the importance of fitted terms in additive models GENERALIZED ADDITIVE MODELS Grasmhrnd Addres Models Mod l Opknra Phera uit Plat M Da iN Eas Pradicion Type mk Serve In l Freios F Sarjan Errors Predict Page New Data Save Enter the name of a data frame to use for computing predictions It must contain the same names as the terms in the right side of the formula for the model If omitted the original data are used for computing predictions Save In Enter the name of an S PLus data frame in which predictions and standard errors are to be saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame This allows you to keep predicted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used
459. prh peat eka z Vansble Opbons 5 s o g F ee Foot Cretthetinn Task F Contra Conmciion Type ofAsek Test Amni Sum Fas Eagrmel Fsik Earn artis F Friii Feasts Data Data Has a Grouping Variable Check this if one column in the data frame is a grouping indicator that categorizes cases into two groups In this case select the response from Response Variable and the indicator from Grouping Variable Data Frame Select a data frame x Variable Select a column as the firs sample when the data frame does not have a grouping indicator y Variable Select another column as the second sample when the data frame does not 390 TWO SAMPLE WILCOXON TEST have a grouping indicator gt Tip Test Hypotheses Options Results You can have a data frame containing a respons variable and a grouping variable or a data frame with two samplesin two columns Type of Rank Test Chose between Rank Sum and Signed Rank The Signed Rank test is not available when D ata H as a Groups Variable is checked Null Hypoth Mean Enter the difference between the assumed population means of Variable x and Variable y Alternative Hypoth Select the alternative hypothesis For example to perform a one sided test against the alternative hypothesis that the mean of Variable x is greater than the mean of Variable y select greater from the list box Use Exact Distribution Check this to use the exact distri
460. print the results of the Kolmogorov Smirnov test in the designated output window Related S PLus language functions ks gof chisq gof qqplot 385 CHAPTER 17 COMPARING SAMPLES ONE SAMPLE CHI SQUARE GOODNESS OF FIT TEST This dialog performs a onesample chi square goodness of fit test of the distribution of a variable against a hypothesized distribution To perform a one sample chi square goodness of fit test Choose Statistics gt Compare Samples gt One Sample gt Chi Square GOF from the main menu T he dialog shown below appears One tampla Chir Squae Comines A Tiit Det Catia Frame Wore bie Opiom Pis ol Clee OoOo O o Cut Pott 0 horace f Hepothesed Hppahasized Dat ioma Fammen ofthe Seleced Dirinisrim m e m Data Data Frame Select a data frame Variables Erdi Deanin Focculis Save Am F Pimi Feie Select the column to which a one sample chi square test will be applied Options No of Classes Enter the number of cells into which the observations are to be allocated according to the hypothesized distribution for the variable If Cut Points are 386 ONE SAMPLE CHI SQUARE GOODNESS OF FIT TEST Hypotheses Parameters of the Selected Distribution Results supplied No of Classes is set to the number of cut points minus 1 Cut Points Enter a vector of cut points for the cells No of Par Est Enter the number of parameters to be estimated from t
461. property dialog as To X Axis M in To Y AxisMin To X O or To Y 0 For details see the description of the Line page of the plot property dialog You can specify a string or column of text to be used as plotting symbols this option also exists for 3D line and scatter plots and polar plots You can create a plot using a column of text as symbols by selecting data and choosing a Text as Symbol button on the plot palette Clicking this button specifies that the text appearing in the z column of the D ata to Plot page of the property dialog is used to label each data point You can select two columns of numbers one column of text and click the Text as Symbol button to create the plot automatically You can open the property dialog and edit options on the Symbol page to change the plot properties Note 324 Unlike a Comment plot you cannot put a box around the text or move the location of the text relative to the plotting point Comment plots also use the position of the data point as the text if no z column is specified For more information see the description of the Symbols page of the Plot Property dialog PLOT PROPERTIES Pie Charts Data to Plot Specs Fills Labels Specify the name or column number of your data under the X column T he column should contain only positive values if negative values are encountered the absolute value of the negative numbers is used in calculating the size of each pie sli
462. pt Files ssc Cancel Help Figure21 3 Opening a script file To run a script from a Script window You can use the Run button or the Run menu option to execute your scripts To run a script w Click the Run button on the Script toolbar or w From the Script menu choose Run To run a portion of a script 1 Select the linesin the script that you want to run 2 Click the Run button on the Script toolbar or choose Run from the Script menu When you run your script file the Script window title changes to the name of the script followed by running W hen the script is stopped or has ended execution the title changes back to program To save a script into a script file 1 Click the Save button on the standard toolbar or from the File 549 CHAPTER 21 WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS menu choose Save or Save As menu items or type CTRL S 2 If you chose Save As or if the Script window had never been saved before a browser window will appear 3 Use the browser to navigate to the folder of your choice and change the File name text field to the file name of your choice for instance savetrees ssc 4 Click on Save to create a new script file named savetrees ssc Save Script File As Save in a samples cl Ez amp my_scriptssc Object Browser sbf S Scripti ssc File name savetrees ssc Save Cancel Figure21 4 Saving a Script file To print a script You ca
463. pt the formula and return to the Cox Regression dialog Example 2 A stratified model using the Lung data frame T his example creates a formula to model survival as a function of age ECOG performance score ph ecog Karnofsky score ph Karno calories consumed at meals meal cal and weight loss wt I oss stratified by sex The variables time and status contain the survival times and event indicator respectively 1 Choose Statistics Survival C ox Proportional H azards from the main menu 2 In the Data Frame field enter or select lung then click the Create Formula button 3 Create the Survival Response by selecting time from the Choose Variable s list and click the Time 1 button Select status from the Choose Variable s list and click the Censoring Indicator button Click the Add Response button 4 Define explanatory terms by selecting sex from the Choose Variable s list and click on the Strata button from the Add Term s group While holding down the CTRL key select age ph ecog ph karno meal cal and wt loss from the Choose Variable s list Click the M ain Effects button from the Add Term s group 5 A survival formula defining a model based on all the variables Stratified by sex is now created Click the OK button to accept the formula and return to the Cox Regression dialog Example 3 A counting process model using the Heart data frame This example creates a formula to model survival as a function of age of
464. r and weight Additional options are High Low Bar Attributes Specify line color and line weight for the high low average bars 317 CHAPTER 12 EDITING PLOT PROPERTIES Histograms and Density Plots Data to Plot Options 318 These plots display the distribution of a single set of data using bars or lines Thehist function is used for the histogram calculation and the density function for the density line calculation Specify a single X vector of data for the distribution Contin Integer Specify Integer if your data are exclusively integers and Continuous if they are not If Integer is specified for non integer data the data values are truncated to integers Output Type Specify the output type Choose C ounts to have the heights of the bars determined by the number of data points falling into each interval The sum of the heights of all of the bars equals the number of observations in the data vector Choose Density to have the height if the bars determined by the relative concentration of data points within each interval The sum of the area beneath the bars is equal to one The general shape of the distribution is independent of the number of bars specified This output type is necessary for density lines to be drawn correctly Choose Frequency to have the height of the bars determined by the fraction of data that falls within each interval The sum of the heights of all of the bars is equal to one Choose Perce
465. r Column to select a column not being plotted If you choose O ther Column specify the name and data set below Column Specify the column to determine the variation of colors D ata Set Specify the data set containing the data column Number of Colors Specify the number of color intervals to be used in the fill A maximum of 64 colors are available for the color range choices 333 CHAPTER 12 EDITING PLOT PROPERTIES Grids 334 Increment M ultiple Specify a multiplier for the increment in a color range A multiplier greater than one results in greater variation in the color bands the higher the Z value For example for a multiplier of 1 1 your surface will be shaded more heavily with the start color and less with the end color Fill Method Choose to fill the surface in Bands contours or Grids Start M iddle End Color Specify the boundary colors for the shading ranges S PLUS interpolates shading between the specified colors Special Colors If the Fill Type is set to Custom Colors you can specify custom colors in the standard Color dialog A maximum of 16 special colors are available Specify the number of custom colors to use in the Number of Colors field Color Column Specify a column of color numbers D ata Set C hoose a data set containing the color numbers X and Y Data Grids Specify the actual number of X and Y grids If you choose Auto S PLUS determines the number of grids from the number of rows and columns in Z
466. r and Black and W hite T hese styles are used to initialize properties of a new graph sheet T hese properties can be changed in the graph sheet property dialog You can also use the Format Apply Style menu option for graph sheets to modify your graph sheet and plots to match a style specification If you choose color the Color Styles dialog appears Basic C olors Specify the User Colors scheme and the Image C olors scheme to be used for the style See O ptions C olor Schemes to edit the color palettes used in these color schemes Line Auto Change Choose whether to have the line style and or color change each time you add a line plot to the same graph Line styles and colors will rotate in the order specified on the Lines page in this dialog Symbol Auto C hange C hoose whether to have the symbol style and or color change each time you add a plot with symbols to the same graph Symbol styles and colors will rotate in the order specified on the Symbols page in this dialog If no rotating is specified the first symbol style and first color specified will be used for the default CHANGING DEFAULTS AND SETTINGS Lines Page Symbols Page Pattern Color Page Fill Color Page Specifying Color Schemes User Colors Page Pie and Area Auto Change Choose whether to have the pattern style pattern color or fill color change for each pie or area in a newly created pie or area chart Pattern styles colors and fill colors will rotate in
467. r creating data summaries performing hypothesis tests and fitting statistical models M any of the dialogs consist of tabbed pages that allow for a complete analysis including model fitting plotting and prediction Each dialog has a corresponding function that is executed using dialog inputs as values for function arguments U sually it is only necessary to fill in the first page of a tabbed dialog to launch a function call T here are several common controls that govern dialog execution OK button Click here to accept the current state of the dialog execute its corresponding function and dismiss the dialog Cancel button Click here to dismiss the dialog without executing its corresponding function T he state of the dialog is not saved Apply button Click here to accept the current state of the dialog execute its corresponding function and keep the dialog open Rollback arrows Click these to move through previous states of the dialog which were saved when OK or Apply were clicked T hese states are not saved between sessions 341 CHAPTER 14 USING THE STATISTICS MENUS AND DIALOGS Dialog Fields 342 Help button Click here to open a help page for the dialog To get help for functions the dialog calls see the help files for the functions listed at the end of dialog help Each dialog is composed of fields that correspond to arguments in an S PLUS function Dialog fields consist of a variety of control types including edi
468. r each confidence level 5 ChooseOK A pair of confidence bounds is computed and drawn for each percentage specified To have no confidence bounds drawn leave the Line Style field set to None for each pair of confidence bounds 315 CHAPTER 12 EDITING PLOT PROPERTIES Error Bar Plots Data to Plot Line Symbol Options 316 For error bar plots specify the names or column numbers of your X column s and Y column s S PLUS automatically computes the mean and standard error and plots a symbol at the mean unless you already have an error column see No error bar calculation below for details Automatic error bar calculation For vertical error bars you need a single X column and a group of Y columns for automatic error bar calculation The X column determines the positions of the bars along the x axis The Y data are used to compute the mean and size of the error bar A series of Y columns can be specified either in a list for example Y1 Y2 Y3 or in a sequence for example Y1 Y3 The mean and error value for each error bar are automatically calculated using each row of y values corresponding to each x A symbol is plotted at the mean and an error bar is drawn at that point For horizontal error bars you need a single Y column and multiple X columns for automatic error bar calculation No error bar calculation If you already have a column containing the standard error for your data you can specify the column und
469. r format Special data formats required in S PLUS are explained in the following section The following table lists each plot type and the required columns Table 8 2 Required columns for different plot types required O optional Plot Type X Y Z W Area Chart Q multiple Bar Chart Q e Bar Grouped Q multiple Bar Stacked QO multiple Bar with Base at Zero Q o Bar with Error Q e Bar with Error Grouped Q multiple Bar H orizontal Q Bar H orizontal Grouped multiple Q Bar H orizontal Stacked multiple Q Bar Chart 3D Q Q Q Box Plot Q e Q Bubble Plot Q e Bubble Color Plot Q e e e Color Plot Q e e Comment Q e Q text Contour Plot Q Contour Plot 3D Q 194 PREPARING DATA FOR GRAPHING Table 8 2 Required columns for different plot types required O optional Contour Filled Q Q Q Contour 3D Filled Q Q Q Density Plot e Dot Plot Q e Error Bar H oriz multiplex Error Bar Vertical e multipley Q Fit Exponential Curvefit QO e Fit Linear Curvefit QO e Fit Log 10 Curvefit O e Fit Log e Curvefit O e Fit Polynomial Curvefit QO Fit Power Q e High D ensity Line Plot Q e High Density Y Zero Q e High D ensity H oriz QO High Low Plot e e high low Histogram o Histogram with D ensity Line e Levels Plot e e Q Line Plot QO e Line with Isolated Points Q Line with Scatter Q Line with Text
470. r more information about the saved object 451 CHAPTER 18 FITTING STATISTICAL MODELS Haaedom ects Anaheis of Vanence Marked Opinia Phera ue Fig Lontasis Auaign Conant io erie Coarse Options page Contrasts Assign Contrast Choose contrasts for the factors by default the Helmert contrasts are assigned to unordered factors and polynomial contrasts are assigned to ordered factors to Variable s Select one or more variables to which the selected contrast in Assign Contrast will be assigned Contrasts This field displays the selection and assignment chosen in Assign Contrast and to Variable s 452 RANDOM EFFECTS ANALYSIS OF VARIANCE Results page Printed Results Saved Results Haedom Ces Anaheis of Vanence Mal chad Opinia Piera ue Fig Frorias Fazy iy Saved Rests Shoe Cupu Ssa in E ARMOA Table V Rieda l Esimajd Cohrs V Fiesa F Ernimasd K Cogit tor Hoal Fao Short Output Check this to print a short summary of the model fit This includes the model formula the AN OVA table for the model and the residual standard error ANOVA Table Check this to print an AN OVA table for each effect Estimated Coefficients Check this to print the estimated coefficients T here are K 1 such coefficients for each K level factor Estimated K Coef for K level Factor Check this to print K coefficients for each K level factor Fitted Values Check this to save the fitted values Resi
471. r to fit plot summarize and do predictions from one dialog with calls to several key S PLUS functions including m print Im summary Im plot Im and predict Im Modifying the The existing statistics dialogs can be modified or new ones can be written Statistics Refer to the chapter on Programming the User Interface in the S PLUS A Programme s Guide for details Dialogs 345 CHAPTER 14 USING THE STATISTICS MENUS AND DIALOGS 346 CREATING AND MANIPULATING DATA New Data O bject 348 Tabulate 349 Merge Two D ata Frames 351 Random Sample G eneration 353 Density Cumulative Probability or Q uantile 355 Random Number Generation 358 Introduction In S Plus data management and generation can be performed through the default data menu In this chapter each of the data menu items is described in detail Refer to chapter 14 Using the Statistics Menus and Dialogs for an explanation of features common to most dialogs 347 CHAPTER 15 CREATING AND MANIPULATING DATA NEW DATA OBJECT This dialog creates a new data frame matrix or vector See the chapter on Using Data windows for an explanation of the differences between these three generally data frames are recommended To create a new data frame matrix or vector Choose Data New Data Object from the main menu The dialog shown below appears Sealed a dsn nhed in renis TATE Cancel Gissi Select from the list and click OK A Data window will
472. racter allowed is the period Names are case sensitive so X and x are different names When pruning the default name is last pruned When shrinking the default name is last shrunken TREE MODELS Tiot Aigmision Morel Fenu int few Cate aie Sas Ae Freeh Gain Typ ecto Predict page New Data Save Enter the name of a data frame containing data at which predictions will be computed Column names must be those that appear in the formula of the tree model fitted on the Model page By default predictions will be computed at the data used to fit the original tree Prediction Type Select one of vector tree or class See the help file for the S PLUS function predict tree for a detailed description of these options Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names If no value is entered no predictions will be returned Related S PLus language functions tree tree control plot tree text tree sSummary tree print tree prune tree shrink tree plot tree sequence predict tree menuTree 479 CHAPTER 18 FITTING STATIST
473. raph Otherwise your x and y data values are interpreted in document units instead of axes units For example an x y location of 1 2 positions the associated picture one inch up from the bottom of the sheet and two inches from the left side of the Graph sheet X Axis Y Axis Choose which x axis and y axis to scale the picture to By 165 CHAPTER 7 EXCHANGING OBJECTS WITH OTHER APPLICATIONS default the plot is scaled to the first x axis and y axis on the graph Note When the property dialog for the inserted picture is displayed the default size of the picture is always displayed in the Original Width and Original H eight fields T hese fields are for information purposes they are not editable Linking Data You can link data from other applications directly to your S PLUS plots This from Other is useful if the source data is likely to change your S PLUS plot can update ee automatically to reflect the changes You can control whether to update the Applications graph automatically or manually To link data from another application to an S PLus plot 1 Select the data in the source application eg M icrosoft Excel 2 Copy the data to the clipboard using C opy from the Edit menu 3 With your S PLUS plot selected and in focus choose Paste from the Edit menu or w Use the drag and drop method by selecting the data from an application and dragging it to the S PLUS plot Editing Links You can control each
474. ration to create a PowerPoint presentation assuming you have Slides PowerPoint installed Automatically 1 Click on the PowerPoint Presentation button on the Standard toolbar You will see the Welcome screen of the PowerPoint Presentation W izard Click N ext 2 You can now add your graphs to the PowerPoint presentation Click the Add Graph button to find your S PLUS graphs and add them to the list for your presentation or from the standard File O pen dialog select one or more graphs to add to the presentation Click N ext to move to the next page of the wizard 3 Click Finish PowerPoint is started and the graphs you chose are inserted as slides in a new PowerPoint presentation They are inserted in the order you specified in the presentation list in the wizard 83 CHAPTER 3 TUTORIAL 84 USING DATA WINDOWS Working with D ata Windows Creating a N ew Data Window D ata O bjects D ata Frames M atrices Vectors Lists Editing a D ata Set The Current D ata Window Selecting Cells Columns and Rows Selecting Cells Extending the Cell Selection Selecting Columns Selecting Rows Using Keyboard and M ouse Shortcuts Using the Go To Call Option Column Names and Column Numbers Column Lists in the D ata Window Entering D ata Entering D ata from the K eyboard Entering D ata from O ther Sources Editing Data Editing Time Series D ata Moving and Copying D ata Inserting and D eleting Cells Rows and Columns Inserting Columns an
475. rawn for each tick mark Number of Ticks Minimum and M aximum T he number of ticks drawn for each variable will be computed within the range specified here M ajor Ticks Specify the length and the line weight of the ticks Position If Out is selected the ticks will be drawn pointing out from the outer rectangle If In is selected the ticks will be drawn inside of the diagonal rectangles If O ff is selected no ticks or tick labels will be drawn Tick Labels Specify the font font size and font color for the tick labels Variable Labels Specify the font font size font color and special attributes of the variable labels shown in the center of the diagonal rectangles An option to use a column of label names is available M atrix Borders Fills Specify the fill color pattern and pattern color and border line style color and weight for each of the rectangles off of the diagonal Box Margins Specify the horizontal and vertical margins between the rectangles in document units Diagonal Borders Fills Specify the fill color pattern pattern color and border linestyle color and weight for each of therectangles off of thediagonal Box Order Choose Table Order to have the first variable in the upper left corner with the diagonal boxes moving towards the lower right Choose Graph Order to have the first variable in the lower left corner with the diagonal boxes moving towards the upper right Bottom Triangle Choose whether to have o
476. re over T his indicates where the commands will be inserted in the script when you release the mouse When you release the mouse the commands will be written into the Script window starting at the line where you dropped the objects Dragging Function Objects into a Script Window 560 If you drag an S PLUS function object from a Browser window onto a Script window the function definition is expanded in the Script window Thisisa very convenient shortcut to edit a function s body If you want to test your changes select Run in the Script menu or click on the Run button and the new function definition is automatically sent to the S PLUS interpreter THE COMMANDS WINDOW THE COMMANDS WINDOW The Commands window enables access to the S PLUS language and provides backward compatibility for users of earlier versions of S PLUS There are some statistical and analytic techniques that are only available through the Commands or Script windows T his section is a basic introduction to the Commands window The Programme s Guide and the Guide to Statistics discuss the use and features of this powerful tool in detail To open the Commands window click on the button shown in Figure 21 9 Figure 21 9 The Commands window button on the Standard toolbar Unlike other windows only one Commands window can be open Users of earlier versions of S PLUS will see the familiar gt prompt T his is the prompt for an S PLUS command line Tab
477. re ready to import the data simply deselect Preview Only and choose OK T he Filter page works exactly as described in the Import File section above EXPORTING DATA SETS EXPORTING DATA SETS W hen you are exporting to most file types typically you only need to specify the data set file name and file type and the data will be exported into a new data file using default settings You can specify your own settings on the O ptions page of the Export D ata dialog All formats that can be imported from can be exported to with the exception of Sigma plot jnb files Formatted ASCII export is available to the extent that fixed width non delimited columns may be output by specifying a null or empty delimiter with the ASCII export option Both the Data Specs and Options pages share a common lower section T hese controls are described below File name Specify the file name and optionally the path of the file you wish to create upon export Save as type Select the type of file you wish to create Within the Export From group are the following options Data Set Specify the name of the data set containing the columns and rows to be exported Columns and Rows Specify a subset of columns and rows to be exported If no columns or rows are specified or the keyword ALL is specified all columns and rows in the 135 CHAPTER 5 IMPORTING AND EXPORTING DATA data set will be included in the file Expert fete Sa
478. ready exists its contents will be overwritten The saved object will have classht est This must be a valid S PLUS object name any combination of FRIEDMAN RANK SUM TEST alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names Print Results Check this to print the results in the designated output window Related S PLus language functions friedman test print htest menuFriedman 399 CHAPTER 17 COMPARING SAMPLES EXACT BINOMIAL TEST Data Test Hypotheses Results 400 This dialog tests hypotheses about the parameter p in a binomial model given the number of successes and the number of trials in the generating experiment To perform an exact binomial test Choose Statistics gt Compare Samples gt Counts and Proportions gt Binomial Test from the main menu T he dialog shown below appears Dna Teel Hypotheses Mo of Succmesnn Hypothesined p DE Pia ai Tiisala Alemana Hppoh ta nrden z Beie Eata E Print Feceaats No of Successes Enter the number of observed successes No of Trials Enter the number of trials Hypothesized p Enter the probability of success to be tested the default is 0 5 Alternative Hypoth Select the alternative hypothesis two sided true parameter is not equal to p default e less true parameter is smaller than
479. ree with all its branches and leaves T his can lead to alarge amount of output Several results that are not in the tree model object can be saved in an S PLUS object Save In Enter the name of an S PLUS data frame in which a part such as predictions and residuals of the analysis is saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame T his allows 475 CHAPTER 18 FITTING STATISTICAL MODELS 476 you to keep predicted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Misclassification Errors Check this to save misclassification errors See the on line help for residuals tree for details Pearson Residuals Check this to save Pearson Residuals See the on line help for residuals tree for details Deviance Residuals Check this to save deviance residuals See the on line help for residuals tree for details Tiot Aigm sion Model Results Pint Pree Shrink Predict Erard Site Emach Tast E Proportional
480. reg fit survreg distributions anova survreg Other related S PLus language functions formula Im solve Surv 525 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS ACF AUTOCOVARIANCE FUNCTION Time Series Data This dialog estimates and displays the autocovariance autocorrelation or partial autocorrelation for a time series See the on line help for acf for computational details See also chapter 20 Creating and Viewing Time Series and chapter 21 Analyzing Time Series in the Guide to Statistics To estimate autocovariance or autocorrelation Choose Statistics gt Time Series gt Auto Covariance Correlation from the main menu T he dialog shown below appears AF Asie Covnnimace Puncine Pile Tima Series Daia pions Cts Frame Estee Type seater core Ltr Tine Bares I Change Mesrarm Lag Oeti Eana TEE d P FH ACF Rora Data Frame Select or enter the name of a data frame having time series as columns Time Series Select the column containing the time series to be analyzed gt Tip Options 526 You can enter the name of a time series object directly in Time Series For example enter I ynx to compute the correlogram for the lynx time series Estimate Type Select an estimate type autocorrelation to estimate the autocorrelation function the default ACF AUTOCOVARIANCE FUNCTION Results autocovariance to estimate the autocovariance function e
481. render than the traditional style graphics Each execution of a dialog will produce a new Graph sheet device by default T his facilitates comparison of plots generated from multiple calls to a dialog or from calls to multiple dialogs To override this action select O ptions Graph Options from the main menu and clear the Create N ew Graphsheet check box in the Statistical Dialogs Plots group D eselecting the new graphsheet option prevents graphsheet proliferation If multiple plots are requested from within a dialog each plot will be created on a separate page of the graphsheet To add subsequent plots from multiple dialog calls as pages of the current graphsheet select O ptions Graph Options from the main menu then select Each Graph from the Auto Pages drop down list in conjunction with deselecting the Create N ew Graphsheet check box On completion of the execution of a dialog function the default or user named Save As object shows up in the O bject Browser For mode objects such as the results from a linear regression right click context menus are available Right click on the model object in the O bject Browser to display the related menu Most menu choices correspond to the tabbed pages from the dialog This allows you to go back and do plotting and prediction for a mode without relaunching the entire dialog An example for an m linear 343 CHAPTER 14 USING THE STATISTICS MENUS AND DIALOGS a Example Data
482. rm 100 would not repeat 50 itemsinto acolumn of length 150 For further information on specific formatting options see section Formatting D ata Windows page 111 INSERTING AND DELETING CELLS ROWS AND COLUMNS DataFrame Marsala ii ieee fw OOO imitim e F Ct Poo lax E maa Tikai T pre Figure 4 5 Thelnsat Column button drop down lig above and dialog bdow To insert rows 1 Select the row you want to have shifted to make room for the new cells If you want to insert multiple rows select the same number of rows as you want to insert 2 Choose Row from the Insert menu and click OK in the Insert Rows dialog ES aaro oust Figure 4 6 Thelnset Rows dialog Deleting Data To delete data in a cell w Move the cursor to the cell and press DELETE The deleted cell is replaced with an N A which denotes a missing value 103 CHAPTER 4 USING DATA WINDOWS To delete data in an entire Data window 1 Click in the upper left hand corner of the D ata window to select all data Press the DELETE key or choose Clear from the Edit or shortcut menu To remove a column 1 Select the columns you want to remove i 2 Click the Remove Column button Or from the D ata menu choose Remove then Column Or from theC olumn Remove dialog choose OK fr Grae Colum Gounst Hewes oluewns eet ancom Count Figure 4 7 The Clear Columns and Remove Columns dialogs Clearin
483. rnels are scaled so the upper and lower quartiles of the kernel are 0 25 and 0 25 when the bandwidth is 1 Larger values of bandwidth make smoother estimates while smaller values make less smooth estimates The default bandwidth is 0 5 Minimum x Enter the minimum value of x at which to compute the estimate T he default isthe minimum x value Maximum x Enter the maximum value of x to which to compute the estimate T he default isthe maximum x value Number of Points Enter the number of points to smooth in the interval Minimum x M aximum x T he default is the number of points in Variable x Use Specified Values for Estimation Check this to allow specification of a vector of x values When this is checked the Vector of x s edit box is enabled and the M inimum x M aximum x and Number of Points edit boxes are disabled Vector of x Values Enter the name of a vector to specify where the kernel estimate is computed For density estimates the default is a sequence the length of Number of Points ranging from Minimum x to M aximum x T he default for regression estimates is x This field is enabled only when Use Specified Values for Estimation is checked Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This object will be saved as a list with components named x and y This must be a valid S PLUS object name any combin
484. robability are disabled Similarly Q uantile isdisabled when ethe of the other two is checked Distribution Select the distribution for the variable Parameters of the When a distribution is chosen fields for those parameters associated with the Selected distribution are enabled Distribution Enter the parameters of the hypothesized distribution For example if the 356 Cauchy distribution is chosen from the Distribution drop down box the Location and Scale are enabled DENSITY CUMULATIVE PROBABILITY OR QUANTILE Results Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed The only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last The default is last pdist Print Results Check here to print output in the designated output window Related programming language functions punif dunif qunif pnorm dnorm qnorm 357 CHAPTER 15 CREATING AND MANIPULATING DATA RANDOM NUMBER GENERATION 358 T his dialog generates random numbers from a specified distribution To generate random numbers Choose Data gt Random Numbers from the m
485. rowser T hen you can click a plot palette button or you can drag and drop plot buttons onto graphs and then drag data onto the plot buttons or you can use the Graph option on the Insert menu For this example we will create a graph using the plot buttons The 2D and 3D Plots buttons are available on the standard toolbar for creating graphs quickly see figure 3 12 When you click on the 2D or 3D Plot buttons a palette of plot buttons appears For a description of each plot move the mouse cursor over each button in the palette A text description of the plot appears in the status bar at the bottom of your screen in addition to the tooltip When a new graph is created using a plot button a Graph sheet is automatically opened in a new window To create a 2D graph using a plot button and the data frame Exenvirn 1 Import the data set Exenvirn sd2 if you have not already done so M inimize or close the O bject Browser if it is open 2 In the Data window select the data columns for OZONE and RADIATION by dragging across the column headers T he order in which columns are selected determines their default plotting order The first column will be X data and the second Y Note that you can CTRL CLICK on column headers to select discontiguous columns or to select columnsin a specific order 3 Click the 2D Plots button to open a palette of available 2D plot types 4 Click the Loess fit button on the palette see figure 3 14 A locally
486. rs of the rows to use For example Species bear use only bears 1 20 use only the first 20 rows Age gt 13 amp Age lt 20 use only teenagers To edit data specifications for multiple plots 1 Doubleclick in the graph area to display the G raph dialog or right click on the graph area and select Plot Summary Page or right click on the graph area and choose Select O bjects from the Format menu 2 In the lower portion of the plot summary page of the graph dialog select the plot you want to edit 3 Make any desired changes in the fields in the upper portion of the dialog 4 Repeat steps 3 and 4 for each plot you want to modify and choose OK If you choose the Select All button in the Plot Summary page any specifications made in the upper portion of the dialog will take effect on all plotsin the graph T his could be useful for example if you wanted to specify the same X column for all of your plots 189 CHAPTER 8 CREATING A GRAPH PLACING MULTIPLE GRAPHS ON A GRAPH SHEET Graphs can be added to an existing Graph sheet using the plot buttons the menus or drag and drop g x E a 2 o amp BS E gt T T g o x E Q 2 A Medium PMA ClAnti CD28 31 3 625 125 0 78 312 A23187 ng ml Calphostin C nM Figure8 7 Multiple graphs on a page Adding a Graph to an Existing Graph Sheet 190 To add a graph by SHIFT clicking on a plot button 1 Open t
487. rty of the object PropertySubcommandN ame is the subcommand name of the property to change To find out the names of the properties of an object you can use the following script guiGet PropertyNames classname Note that all objects have two properties that may or may not be displayed on the dialog TXPROP_Obj ect Name Subcommand name NewN ame always in position 0 but usually not displayed in a dialog TXPROP_Obj ect Pos ndex Subcommand name N ewlndex always in position 1 but usually not displayed in a dialog To find out the argument names of the properties of an object you can use the following script gui Get ArgumentNames classname The argument names are usually very similar to the corresponding prompts so that figuring out which dialog field corresponds to which property should not bea problem Classinfo objects can be created and modified either using a script or interactively in the browser An example of an script for creating a Classi nfo object is Create ClassInfo Name lm ContextMenu Im DoubleClickAction tabSummar y m DialogHeader Linear Model Object PromptList 0 0bject Name To create a Classinfo object in the Browser first create a page that displays these objects Then right click on the Classinfo root node in the left pane of the O bject Browser and choose Create C lassi nfo CUSTOMIZING YOUR S PLus SESSION Overview Overview 597 Ch
488. ry of the model fit in the designated output window Save In Enter the name of an S PLUS data frame in which fitted values weights and residuals are is saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame This allows you to keep predicted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Fitted Values Check this to save the fitted values from the model in the object specified in Save In Residuals Check this to save the residuals values from the model in the object specified in Saveln LTS Weights 0 s 1 s Check this to save weights with a value of 1 for observations having reasonably small residuals and a value of 0 for observations having large residuals These weights can later be used in an ordinary least squares regression Robust Regresan Fl Ei Monal Optar Fira uta Fiat Pins Ophore fF Renidues wa Fi Numer of Erara Ponte To iiher M Fiasko sa idos P F Rendina Homa G P Amuda Robis Diriarce
489. s Predict Page New Data Enter the name of a matrix or data frame containing columns with the same names as those used in the original analysis Save Save In Enter the name of an S PLUS data frame in which predictions from the analysis are saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame This allows you to keep predicted values from a model with the original data or to keep the predictions from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Predictions Check this to save predictions to the data frame specified in Save In Related S PLus language functions for Factor Analysis factanal factanal object factanal fit mle 498 FACTOR ANALYSIS factanal fit principal factanal mle control factanal start mle predict factanal fitted factanal rotate factanal biplot factana Other related S PLus language functions princomp 499 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS PRINCIPAL COMPONENTS ANALYSIS This dialog calculates the principal comp
490. s Formatting Polar Axes Adding M ulti line T ext Adding Special Characters and Formatting T ext Adding Titles and Legends Adding 2D Axis Titles Adding 3D Axes Titles Adding aD ate and T ime Stamp Adding aLegend Formatting the Legend Items Adding Labels for Points Adding a Curve Fit Equation Adding Lines Shapes and Symbols Summary of the Annotation Palette Chapter 11 Working with Graph Objects Graphic O bjects Selecting O bjects Summary of Format and View M enus M oving and Copying O bjects U sing D rag and D rop Using Cut Copy and Paste O ptions Sizing O bjects Editing O bjects 226 227 228 233 233 234 238 239 244 249 250 251 254 256 257 259 261 265 267 268 268 269 270 271 273 274 275 277 281 281 282 284 288 288 288 289 290 CONTENTS Arranging O bjects on the Graph 292 Overlapping O bjects 292 Aligning O bjects 292 Aligning O bjects U sing Snap to G rid 293 Distributing O bjects 293 Chapter 12 Editing Plot Properties 295 Plot T ypes 295 Changing the Plot T ype 295 Plot Properties 299 Common Plot Properties 300 Area C harts 303 Bar Charts V ertical and H orizontal 2D 304 Box Plots 307 Comment Plots 309 Contour Level Plots 2D 311 Curve Fitting Regression 313 Error Bar Plots 316 H igh Low C lose Plots 317 Histograms and D ensity Plots 318 Line amp Scatter Plots 2D 319 Pie Charts 325 Polar Plots 326 Vector Plots 327 Scatterplot M atrix 328 Contour Plots 3D 3
491. s See the on line help for ari ma diag plot for details on the different plot components AAI Modeling SE Mode Fig Optane Chagnontce Forta Farmmeiers Tim Paida To Freee P Piotledudiag Standard Errors Seen Ms inoaii vaian Opona Lad ee a ll 1e Parameters Time Periods To Forecast Enter the number of time periods to forecast beyond the end of the series T he default is 5 Innovations variance Optional Enter the estimated innovations variance if different than the concentrated prediction error variance computed from the modal Plot Including Standard Errors Check this if a plot is desired Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of ARIMA MODELING alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names T he saved object will be a list with two components the estimated mean of the forecasts and the approximate forecast error This is only saved when a string is entered in the Save As field Related S PLus language functions acf ar arima mle arima diag ari ma diag plot arima forecast menuAri ma 533 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS
492. s Enter the maximum number of iterations permitted by the optimizer in computing the parameters of the model T he default is 15 Likelihood Evals Enter the maximum number of times that the likelihood should be evaluated T he default is 30 530 ARIMA MODELING AHIRA Modelni EEE Ping Opinna Ciagn omic Forca Save See All P Aicorakion of Fed aisha P Pia Disgo CV Fotmanes mirit F Reidas l Siiatertiged Aairiiaii Diagnostics Page Save Save As Enter the name for the object in which to save the results of the analysis See the on line help for ari ma di ag to find out more about the contents of this object If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names Autocorrelation of Residuals Check this to save the autocorrelation of the residuals Portmanteau Statistic Check this to save a list representing the Portmanteau goodness of fit statistic Residuals Check this to save residuals 531 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS 532 Forecast Page Standardized Residuals Check this to save the standardized residuals Plot Diagnostics Check this to generate a standard plot of diagnostic
493. s more likely to encounter them as the results of some computation The example matrix cereal attitude gives the percentage of people agreeing with 11 statements such as Reasonably Priced about 8 brands of cereals It has 11 rows and 8 columns For an extensive discussion of matrices and their uses see the chapter on D ata O bjects in the Programme s Guide In S PLUS a vector is an ordered set of elements having the same mode Commonly used modes are character numeric complex and logical Like the rows and columns of matrices and data frames the elements of a vector can have names Each column in a data frame is a vector The user will also encounter vectors as the result of some computations The example vector ozone median gives 41 median ozone readings taken over time For an extensive discussion of vectors and ther uses see the chapter on Data O bjects in the Programme s Guide Lists are collections of other objects T heir components can be data frames matrices vectors other lists functions or any other S PLUS objects Lists are used to contain related data objects such as computational results from a linear regression fit The example list evap has a component x which is a matrix of independent variables and a component y which isa vector of daily evaporation amounts EDITING A DATA SET For an extensive discussion of lists and their uses see the chapter on Data O bjects in the Programme s Guide Editin
494. s of M enultem objects T his section shows how to work with menus through the user interface W hile it is not hard to create or modify menus through the user interface as shown in this section it is sometimes easier to work with menus using the S PLUS programming language As you work with menus by the methods in this section take an occasional moment to review the program code generated in the history log See the S PLUS Programme s Guide for a systematic explanation of the program code To create a menu item first open the Object Browser and filter by M enultem to see the hierarchy of menu items N avigate to the menu item above which the new menu item should appear Right click on this menu item and select Insert Menultem from the context menu The property dialog shown in figure 22 15 appears Figure22 15 The property dialog for a M enultem object M enultem page 586 To modify the default settings that appear in this property dialog in the future right click on the default object icon labeled M enultem choose M enultem or Command from the context menu fill out the dialog with the desired default values and click OK Name Enter the name of the M enultem object MENUS Type Select the type of M enultem object Menu creates a submenu e Menultem causes an action to occur when selected e Separator displays a horizontal bar in the menu visually separating two group of menu items Document Type Dep
495. s of the analysis If an object with this name already exists its contents will be overwritten The saved object will have classht est This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed The only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names FISHER S EXACT TEST Print Results Check this to print the results in the designated output window Related S PLus Language Functions fisher test print htest menuFisher 405 CHAPTER 17 COMPARING SAMPLES MCNEMAR S CHI SQUARE TEST 406 Data Results This dialog performs McNemar s chi square test on a two dimensional contingency table defined by two classification variables in a data frame To perform McNemar s chi square test Choose Statistics gt Compare Samples gt Counts and Proportions gt McNemar s Chi Square from the main menu T he dialog shown below appears KicNemers Chi S quae Test BE pes F Appir Conanuny Comecion Cisia Preme Presale Cites fhnshon Variable Ose Bae Ai E E Friml Fiais Dias Iiostion variae Ta Data Frame Select a data frame Classification Variable One Select the factor column that contains the first classification variable It must have at least 2 levels Classification Variable Two Select the factor column that contains the other classification variable This variable
496. s of variance tree models and more N ow let s fit a tree based model 1 Close all Data windows and Graph sheets and restore the O bject Browser window 2 Click once on fuel frame in the left pane of the Object Browser then select the M ileage column for the response variable simply by clicking on it Then CTRL CLICK to select Weight Disp and Type as the predictor variables 63 CHAPTER 3 TUTORIAL 3 Choose Statistics Tree M odels The Tree Regression dialog appears with the Data Frame field filled in The response and predictor variables are written out in the Formula field 4 Click OK A tree based model appears in a Graph sheet and summary statistics for the model are displayed in the Report window We see that Displacement is the most important predictor of M ileage followed by Weight and Type IE ibe Lal a badal Ran r Fia Purasia Pradet Das Fitre Diaa ss fua ieira bin Fic of Cle Mieka Spi Wega I E Subesi iana Hin Rocke Sine fit FF rad fisar sah Maung Vakan Min Foia Deviemes E ave hodh Ghia Srania iia F orma i Farala Mileage weight e Dep Tape Create Pearl Ca eej aE e Figure3 11 The Tree Regression dialog notein particular the filled in Formula box Summary With S PLUS you can interactively explore analyze and visualize your data with unsurpassed ease of use and flexibility 64 IN DEPTH TOUR IN DEPTH TOUR S PLUS lets you import and export data fr
497. sample generation Choose Data gt Random Sample from the main menu The dialog shown below appears Raeadom samplo Genpreteon EE Bampira Opbn rea Dais l l Semple ath Replacement Sampleg Peak F Ei Daad wii Senpi Sire Premie Ean Ai issisarmpis l Priel Reseke Data Select a data frame from the drop down box or enter an expression for a vector For example enter either 100 or 1 100 to sample from the integers 1 through 100 Sampling Prob Select the variable containing sampling probabilities or type in an expression for a vector of probabilities T he default is equal probability Sample Size Enter the size of sample T he default is the length of the vector or the number of rows in the data frame T o perform a random permutation of the data use this default and uncheck the Sample with Replacement field Sample with Replacement Check hereto sample with replacement If this box is unchecked once a value has been sampled it will not appear again If it is checked the same value can be selected twice Set Seed with 353 CHAPTER 15 CREATING AND MANIPULATING DATA 354 Results Enter arandom number seed used in therandom generation algorithm W hen this field has a value rerunning the dialog in the same state will reproduce the same data Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritt
498. se controls are described below File Name Specify the file name and optionally the path of the file containing data to import Files of type Select the format of the file you wish to import Within the Import To group are the following options Data Frame Specify the data frame to contain the imported data You can specify the name of an existing data frame or a new data frame If the specified data frame does not already exist S PLUS automatically creates it If the specified data frame does exist you will be warned that you could overwrite data By default the imported data will automatically be displayed in a D ata window You can choose to turn this automatic display feature off using the D isplay in Grid option in the O ptions G eneral Settings D ata Import menu Target Start Col Specify the starting column for the imported data T he start column can be any column in an existing data frame or anew data frame Preview Only Choose Preview O nly to have the data read entirely as character columns into a data frame called PREVIEW This lets you view all of your data regardless of the data types When you are ready to import the data simply deselect Preview Only and choose OK If no Name Row see Options page in Figure 5 1 is specified when previewing a spreadsheet file then spreadsheet 119 120 CHAPTER 5 IMPORTING AND EXPORTING DATA style letters e g A AB will automatically be created and used as column names
499. se the conditioning variables specified on the graph s multipanel page Select N one to use the full data set for this plot in each panel no conditioning is used This is useful if you would like to draw a reference plot in each panel Select Specified Columns if you would like to override the specifications for the conditioning variables for this plot D ata Set If Type is set to Specified C olumns specify the name of the data set containing the column to use for conditioning Column s If Type is set to Specified Columns specify the variable name to use for conditioning Draw in Panels Select All to have the plot drawn in all panels or select a panel number to have the plot drawn only in asingle panel 211 CHAPTER 9 WORKING WITH TRELLIS GRAPHICS Formatting Plots within Panels Formating Panels Fill Border Page Multipanel Page 212 W henever a plot is placed on a Trellis graph a plot panel object is created for each panel They are contained by the plot in the O bject Browser and their properties can be changed by right clicking on their icon T hese panel objects have display attributes of their own that can be used to override the plot s display attributes For example you might want to set the line color of the plot to green in the third panel and to blue in the rest of the panels To edit individual panels 1 Open the object browser Right click in the window and select GraphSheet 2 Expand the branc
500. select a range of contiguous rows w Drag the mouse across the desired row numbers To select a range of discontiguous rows w cTaL click in the row number for each row To modify the row name w Doubleclick on a cell of the right grey column 91 CHAPTER 4 USING DATA WINDOWS Using Keyboard and Mouse Shortcuts Table 4 2 Keyboard and mous shortcuts The following table outlines the effect of the direction keys and mouse movements in S PLUS D ata windows CTRL Page D own CTRL Spacebar 92 column M oves to last row current column Selects a column Keyboard Action Mouse Right Arrow Selects the cell to the right Click the cell to the right Left Arrow Selects the cell to the left Click the cell to the left Up Arrow Selects the cell above Click the cell above Down Arrow Selects the cell below Click the cell below Page Up M oves the screen up Click up scroll bar arrow Page Down M oves the screen down Click down scroll bar arrow CTRL Left M oves the screen left Click left scroll bar arrow CTRL Right M oves the screen right Click right scroll bar arrow CTRL Home M oves to first column first row D rag sliders to up and left arrows CTRL End M oves to last column last row D rag sliders to down and right arrows Home M oves to first column same row D rag horizontal slider to left arrow End M oves to last column same row D rag horizontal slider to right arrow CTRL Page
501. side the graph to display the Graph Properties dialog Click on the Position Size page For Plot Origin Position specify X and Y values and choose OK To format the plot area 1 Doubleclick inside the graph to display the Graph Properties dialog 2 Click on the Fill Border page M ake the desired formatting changes to the plot area border and fill 3 ChooseOK In the Graph Properties dialog the Position Size page has the following options Graph Position T he graph area is positioned according to the horizontal and vertical specifications entered here The distance is measured from the bottom left hand corner of the page area to the lower left hand corner of the graph area Graph Size Specify the horizontal and vertical lengths of the graph area Graph Interior Margin Specify amargin for automatic plot area calculations within the graph area border To the extent possible axes labels will not be drawn within this area Plot Origin Position T he plot area is positioned according to the horizontal and vertical specifications entered here T he distance is measured from the lower left hand corner of the graph area to the lower left hand corner of the plot area If Auto is specified the plot area will be repositioned to allow room for axes labels Plot D isplay Size Specify the horizontal and vertical lengths of the plot area If Auto is specified the plot area will be resized to allow room for axes labels 235 CHAPTER 10 FOR
502. siduals Save In Enter the name of an S PLUS data frame in which a part such as fitted values and residuals of the analysis is saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the 491 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS 492 name of a data frame that already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame T his allows you to keep fitted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Related S PLus language functions manova Summary manova coef dummy coef aov raov Im FACTOR ANALYSIS FACTOR ANALYSIS This dialog performs a factor analysis on a set of observations of many variables See also chapter 17 of the Guide to Statistics page 483 To perform factor analysis Choose Statistics gt Multivariate gt Factor Analysis from the main menu The dialog shown below appears Factor Analysis x Model Options l Results Plots Predict r Data m Model Data Frame lethanol v Number of Factors fi Subset Rows with Method
503. sing the source application s menus and toolbars 3 When finished click anywhere outside the embedded object to return to S PLUS s menus and toolbars W hen the server program source application for the embedded object does not support in place activation the object is created edited and displayed in the server program EMBEDDING OBJECTS FROM OTHER APPLICATIONS Creating from Existing Files Displaying as Icons Formatting and Placement To edit the embedded object in the server program 1 In your S PLUS Graph sheet double click the embedded object or Select the object choose O bject from the Edit menu then choose O pen 2 A separate window opens within the server program Create and edit the object using the source application s menus and toolbars 3 When finished choose Exit and Return from the server application s File menu If you want to update the embedded object in the S PLUS graph choose U pdate from the File menu You can embed existing files as objects in your S PLUS Graph sheets The original file does not remain linked to the embedded file You can edit the embedded file without changing the original file To create an embedded object from an existing file 1 With a Graph sheet in focus choose O bject from the Insert menu 2 To embed an existing file choose the Create from File button and select the file name 3 ChooseOK to embed the file In S PLUS you can also insert pictures into your
504. sion must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the moda If this box is not checked S PLUS will report an error and halt the routine if any row is found to have a missing value in any of the terms in the modal Formula Enter a formula specifying the desired model Create Formula Click this to open a formula builder dialog used to construct a formula specifying the desired model See the chapter Building Formulas for more information Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names T he default is last raov See the on line help for aov object fo
505. sp fits a regression model with Fuel as the response and Weight and Disp as predictors For more information on formulas see the chapter on Building Formulas Create Formula Click this to open a formula builder dialog used to construct a formula specifying the desired model See the chapter Building Formulas for more information Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names T he default is last Itsreg T he saved object will have class I t s See the on line help forl ts obj ect 429 CHAPTER 18 FITTING STATISTICAL MODELS for more information about the saved object Options Page Number of Residuals in SS to Minimize Enter the number of squared residuals whose sum will be minimized T he default isf oor n p 1 2 wheren is the number of observations and p isthe number of predictors in the model Results Page Printed Results Short Output Check this to display a short summary of the model fit in the designated output window 430 ROBUST LTS REGRESSION Saved Results Long Output Check this to display a more detailed summa
506. ssion dialog fits logistic regression models using a generalized linear model It calls the same functions as the Generalized Linear M oda dialog with the family restricted to Binomial and the default link set to logit See page 456 for details To perform logistic regression Choose Statistics gt Regression gt Logistic from the main menu LOG LINEAR REGRESSION LOG LINEAR REGRESSION The Log linear Regression dialog fits log linear regression models using a generalized linear model It calls the same functions as the Generalized Linear M oda dialog with the family restricted to Poisson and the default link set to log See page 456 for details To perform log linear regression Choose Statistics gt Regression gt Log linear from the main menu 427 CHAPTER 18 FITTING STATISTICAL MODELS ROBUST LTS REGRESSION This dialog fits robust linear regression models via least trimmed squares LTS It calls the Itsreg function and its print summary and plot methods To perform robust regression Choose Statistics gt Regression gt Robust LTS from the main menu T he dialog shown below appears Ath Pegres mon Mazal piara Pena Fie frog hts E Subest Aran wih Se Model Dhyan F Oma Roe ih hing Vaks Coat hi Formula Fomu P Chais Fommil Data Data Frame Select a data frame gt Tip 428 You can typeinto the D ata Frame edit box any expression which evaluates to a data
507. ssion plane Regression plots are a special case of 3D line and scatter plots You can automatically create a regression plot by selecting data and choosing one of the Regression buttons on the plot palette Clicking one of these buttons specifies the D raw Plane option on the Regression page of the plot property dialog Typically the Line and Symbol Styles are set to None when a regression plane is being drawn Surface plots can use gridded and irregular data Gridded data can be specified in rectangular matrix form or stacked where the Z data is given in one long column and X Y and Z values are given in ascending order Irregular data is specified in a rectangular matrix form You will get better results if the data are distributed fairly uniformly in X and Y and do not contain sharp spikes or drops in Z If only X and Y are specified and they are of equal lengths a2D histogram will be computed for the Z values See section Preparing Data for Graphing page 194 for details on specifying gridded and irregular data For 3D surface plots you can specify line style and weight and cropping information Additional options are Draw Mesh Choose whether to draw the surface mesh T his is useful when you want a color filled surface without a mesh drawn Mesh Attributes Specify the line style top color bottom color and line weight for the surface mesh T he bottom is the underside of the surface the top is the upper sid
508. st dialogs 379 CHAPTER 17 COMPARING SAMPLES ONE SAMPLE T TEST 380 Data Hypotheses Confidence Interval Results T his dialog performs a one sample t test on data held in a data frame To perform one sample t test Choose Statistics gt Compare Samples gt One Sample gt t Test The dialog shown below appears Dnt Corkd ance intereal Diets Frame Confidence Loved iE watalia OS m Hypotheses Save Aa ZE Malha heen o o Fee Atemete Hypoh inse Data Frame Select a data frame Variable Select the column to which the one sample t test will be applied Null Hypoth Mean Enter the assumed population mean of Variable Alternative Hypoth Select the alternative hypothesis For example to perform a one sided test against the alternative hypothesis that the mean of Variable is greater than the null hypothesis mean select greater Confidence Level Enter anumber between 0 and 1 to be used as the confidence level Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of ONE SAMPLE T TEST alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults
509. stribute menu choose Vertical or H orizontal The selected objects are placed equal distances apart on the Graph sheet between the two outermost selected objects Saving graph object defaults is useful if you need to create multiple objects with similar properties To save the defaults for the selected object 1 Right click on the object 2 From the shortcut menu choose Save O bject as D efault 3 You could also go to the Options menu and select Save O bject As D efault W hen you save the current settings as defaults the new defaults will affect all 293 CHAPTER 11 WORKING WITH GRAPH OBJECTS Deleting Objects 294 new objects of that type created from that point on in the current session and in future S PLUS sessions O bjects created using previously defined defaults will not be affected by the default changes O bjects can be deleted from a graph at any time To delete an object 1 Select the object or objects you wish to delete 2 From the Edit menu choose Clear or press the DELETE key Once an object is deleted you may undo the deletion immediately From the Edit menu choose Undo and the item is restored or click on the Undo button O bjects can be removed from the Graph sheet but not permanently deleted by using the Cut command T he object is placed on the Clipboard so that it can be pasted in another location on the current Graph sheet or in another document EDITING PLOT PROPERTIES Plot Types
510. t boxes drop down lists multi select lists check boxes and radio buttons Some fields are required while others are optional Fields corresponding to optional arguments generally contain appropriate defaults these may be modified Some of the fields common to many dialogs are described in the following paragraphs M any dialogs include a Data Frame field This will be automatically filled with the name of the data frame you have selected in the object browser or if you have a data editor window open and have selected any values in it the name of that data frame To specify another data frame select another one from the drop down list the listed data frames are limited to those that have been filtered by an object browser Alternatively directly type the name of a data frame or any S PLUS expression that evaluates to a data frame in the field Some dialogs have a Variables field T his will be automatically filled with the names of the columns of the data frame you have selected in the object browser or data editor Some dialogs require a Formula T his will be automatically filled if you have selected columns of a data frame in the object browser or data editor The first selected column is the response and the remaining columns are the predictors Some dialogs such as Survival and N onlinear Regression require a special formula and are not automatically filled in If you do not want the formula that automatically appears you can type a
511. t attributes for the axes title text see section Adding 3D Axes Titles page 268 Specifications made here apply to the entire line of title text If you want to make specifications on a character by character basis you can use in text formatting codes when editing your text in the Text field Axis Minimum and Maximum Specify the minimum and maximum values for thex y and z axis Choose Auto to have the minimum and maximum set equal to the minimum and maximum values in the data Tick Interval Specify the tick interval between tick marks Number Size Specify whether the tick interval should be interpreted as the number of ticks or the distance between ticks 249 CHAPTER 10 FORMATTING A GRAPH Formatting 3D The formatting for the X Y and Z axes labels can be specified in the 3D Axes Axes Labels X Text Y Text and Z Text 250 dialog In the 3D Axes dialog the X Text Y Text and Z Text pages have the following labeling options for each 3D axis Label Type See the table of Label Types in section Formatting 2D Axis Labels page 244 for a complete list of display type options Precision Specify the number of decimal places to display after the decimal for numeric labels Column If you choose Column for the Label Type specify the name of a column containing the labels here The first row in the column is used to mark the first major tick mark the second row the second tick mark and so on You can leave any number
512. t be moved and Axes Titles sized interactively The text for 3D axes titles is specified from within the 3D Axes dialog and cannot be multi line H owever you can add multi line text to 3D graphs in the form of comments and titles 268 ADDING TITLES AND LEGENDS To add 3D axes titles 1 or 1 2 D ouble click the axes or select the axes and choose Selected 3D Axes from the Format menu From the 3D Axes dialog choose the X Text Y Text or Z Text page M ake the desired changes for the Text Font Size and Color fields and choose OK From the Insert menu choose Title then Axis The3D Axes dialog appears for editing See section Adding Special Characters and Formatting Text page 265 for more information on specifying text Adding a Date A date amp time stamp lets you display the date and time along with specified and Time text You can format the text using the in text codes described in the section Adding Special C haracters and Formatting Text page 265 Stamp See section Adding Multi line Text page 261 for information on editing and formatting existing text To add a date stamp 1 or 2 Click the Date Stamp tool on the Annotation Palette The mouse pointer will change to represent the D ate Stamp tool In the Graph sheet click and drag the Date Stamp tool until you have a text box of the desired size Release the mouse button A default date stamp is placed on the graph The text box
513. t line when it encounters a data point symbol T hisis useful when you do not want the plot line to bisect the symbol If this option is not selected the plot line is drawn through each symbol Symbol Style Specify the symbol style to be used for plot symbols and extra symbols Symbol Color Specify the color for symbols 301 CHAPTER 12 EDITING PLOT PROPERTIES Fills By Item Fill and Border Specifications 302 Symbol H eight Specify the height of asymbol in inches cm Symbol Line Weight Specify the line weight for outlining symbols Symbol Frequency Specify how frequently symbols will be displayed on data points To place a symbol at each data point enter a 1 a value of 3 indicates that every third point will be plotted with the symbol If you choose 0 no symbols will be plotted Jitter Symbols Specify to introduce random noise in a particular direction to the data plotted Jitter Factor Specify the magnitude of noise introduced to the data plotted You can choose a solid color for each item to be filled fill the iten with a series of shades determined by a range of colors or specify new colors for the filled regions using the standard color dialog Fill Type Choose By Item the title of the By Item option varies depending on the plot type to use the fill colors from the By Item page of the dialog otherwise they are ignored Choose one of the Color Range options to have S PLUS interpolate between
514. t menu right click menu for this object in the browser This is the name of a Menultem of type M enu which must have been defined in the standard way for menus DoubleClickAction this contains the name of a Menultem of type Menultem that is it is a single item instead of an entire menu or afunction T his specifies the action that will happen when the user double clicks on the object in the browser It allows a function to be called when the user double clicks Show Dialog On Run if enabled the dialog for the M enultem or function will be displayed before execution DialogH eader this allows the dialog header for the associated object to be defined T his is only useful for interface objects 595 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE Creating and Modifying a ClassInfo Object 596 PromptList this allows dialog prompts to be specified and overridden T he syntax is the same as it is for the corresponding property of Functionlnfo objects 0 amp My Prompt 2 Another amp Prompt PropertySubcommandName L amp ast Prompt That is it is alist of assignments in which the left hand side denotes the property whose prompt is going to be overridden and the right hand side denotes the new prompt T here are two ways of denoting the property by position starting with 0 with the number preceded by a and by property subcommand name In the example above 0 denotes the oth prope
515. t movement You can turn off S PLUS s smart cursor using the General Settings dialog under O ptions If you press up or Down the data is entered and the cursor moves in the appropriate direction T he LEFT and RIGHT direction keys enter the value and move in the appropriate direction if pressed when the cursor is at the far left or far right of the contents of the cell If you begin typing in a cell that already contains a value the old value is overwritten You can use this option to go to a specific cell location in a D ata window To use the Go To Cell option 1 From the View menu select Go To Call 93 CHAPTER 4 USING DATA WINDOWS 2 Specify the column name or column number and row number of the cell to go to 3 ChooseOK T he cursor will be positioned at the specified cell location If you want to extend your selection from the active cell to the location entered in the Go To Cell dialog hold down the sHirt key while choosing OK For example if column 1 row 5 is the active cell and you specify column 5 row 5 in the Go To Cell dialog and press sH FT oK the selection is extended from column 1 row 5 to column 5 row 5 cou ARENT Figure4 1 TheGo To Cal dialog Column Names A column isa vertical group of data cells typically containing the data for a and Column Numbers 94 given variable S PLUS is column oriented meaning that most of the operations work on columns as units Each column i
516. t results in the designated output window Related programming language functions runif rnorm rexp 359 CHAPTER 15 CREATING AND MANIPULATING DATA 360 SUMMARIZING AND EXPLORING DATA Data Summaries Summary Statistics 362 Contingency Table 365 Correlations and C ovariances 368 Smoothing Local Regression LO ESS Smoothing 370 Supersmoother 372 Kernel Smoother 374 Spline Smoother 377 T his chapter describes some dialogs to summarize data T hese include dialogs for computing mean standard deviation correlation contingency tables and so on Users can plot data and get smooth curves to visualize relationships between variables Refer to chapter 14 Using the Statistics Menus and Dialogs for an explanation of features common to most dialogs 361 CHAPTER 16 SUMMARIZING AND EXPLORING DATA SUMMARY STATISTICS T his dialog provides basic summary statistics for a data frame matrix or vec tor of data To generate summary statistics Choose Statistics gt Data Summaries gt Summary Statistics from the main menu T he dialog shown below appears Summa Ss baer tec arable an P tiin Sura by Group E Thee usis F Memu E Hube of Flown E Huria of Hirang Areas l beca F 48 Deetion b Sa mee Parsa FF Gummy Diei Variables FF Pree Feats Eaves Se et e wal Data Data Frame Select a data frame Variables Select the columns of the data frame for w
517. t the alternative hypothesis For example to perform a one sided test against the alternative hypothesis that the mean of Variable x is greater than the mean of Variable y select greater from the list box Confidence Level Enter anumber between 0 and 1 to be used as the confidence level Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names W here appropriate Save As defaults to a name that starts with last For example last Im is the most recent linear regression model fit Print Results Check this to print the results of the two sample t test in the designated output window Related S PLus language functions t test 389 CHAPTER 17 COMPARING SAMPLES TWO SAMPLE WILCOXON TEST T his dialog performs a two sample Wilcoxon Rank Sum test or a Wilcoxon Signed Rank test To perform two sample Wilcoxon test Choose Statistics gt Compare Samples gt Two Samples gt Wilcoxon Rank Test from the main menu T he dialog shown below appears Twa sampis Wilcoxon Tari Bee Data Hepat sea M ms Hiss a Grouping warisi Mutts Meem 0 Cus Fiame Ahama Hy
518. t to be formatted 3 Choosea font size from the toolbar s font size dropdown list Formatting Text To format text as bold italic or underline as Bold Italic or Underline 1 Open atext edit box and add text 2 Highlight the text to be formatted 3 Click the Bold Italic or Underline toolbar button Formatting To format text as superscript or subscript Superscripts and i Subscripts 1 Open a text edit box and add text 2 Highlight the text to be formatted 3 Click the Superscript or Subscript toolbar button or 4 Right click on the edit box and select Superscript or Subscript from the menu Adding Special You can specify text in Symbol font Greek and Extended ASCII characters Characters Using using the Symbol dialog a Dialog To insert special characters using the Symbol dialog 1 Open atext edit box 2 Right click on the edit box and select Symbol from the menu Special characters are available in the Symbol dialog 265 CHAPTER 10 FORMATTING A GRAPH 3 Click on asymbol to select it and click Insert to add the symbol to the edit box without closing the dialog To close the dialog select Close 266 ADDING TITLES AND LEGENDS ADDING TITLES AND LEGENDS Inserting a title is different from inserting regular text because a title is positioned automatically See section Adding Multi line Text page 261 for information on editing and formatting titles To add a main title or a subtitle 1 From th
519. t will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of 436 LOCAL REGRESSION Plot Page Plots rows as the number of observations used in the model fit then the saved values are appended to this data frame T his allows you to keep fitted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Local Regitsibn ll 23 Mode Optore Resu Ar Predict Plots pda ri E A ra wa Fi T Wechiche Erea Sige Apniduse a F V inchiuds Asgpist Aospores a Fi Numiser of Extreme Points To beder F Aaii Nima OD d r Aenda Fi Soren r Gord Fiji of Fitted sg Frede Residuals vs Fit Check this to display a plot of the residuals versus the fitted values Sart Abs Residuals vs Fit Check this to display a plot of the square root of the absolute values of the residuals versus the fitted values This plot is useful for checking for the constant variance assumption of the model Response vs Fit Check this to display a plot of the response variable versus the fitted values Theline y x is also drawn on the graph Residuals Normal QQ Check this to display aN ormal quantile quantile plot of the residuals Residual Fit Spread C
520. ta g SbstAomh o Sewe Model Dba F Omfg woth hieng vale Surette aniline Firmu Forres Crees Foemels Model Page Data Data Frame Select a data frame gt Tip You can type into the D ata Frame edit box any expression which evaluates to a data frame 506 NONPARAMETRIC SURVIVAL Formula Model Save Model Object Weights Enter the column that specifies weights to be applied to all observations used in the survival model To weight all rows equally leave this blank Subset Rows with Enter an S PLUS expression which identifies the rows to use in the analysis To use all the rows in the data frame leave this field blank The expression must evaluate to a vector of logical values TRUE values are used FALSE values are dropped or a vector of indices identifying the numbers of the rows to use Examples Species bear only bears are used 1 20 only the first 20 rows of the data are used Age gt 13 amp Age lt 20 only teenagers are used For more information on constructing logical expressions see the S PLUS Programmer s Guide Omit Rows with Missing Values Check this box to omit from the analysis any rows in the data frame that contain missing values for any of the variables in the model If this box is not checked S PLUS will report an error and halt the routine if any row is found to havea missing value in any of the terms in the modal Formula Enter a formula specify
521. te variances and covariances Correlation Check this to generate correlations CORRELATIONS AND COVARIANCES Results Fraction to Trim Enter anumber between 0 and 0 5 which is the proportion of cases to trim before calculation of the correlations Save As Enter the name for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed T he only non alphanumeric character allowed is the period Names are case sensitive so X and x are different names Print Results Check this to print the results in the designated output window Related S Plus language functions cor var 369 CHAPTER 16 SUMMARIZNG AND EXPLORING DATA LOCAL REGRESSION LOESS SMOOTHING This dialog performs a locally weighted regression smooth using k nearest neighbors at each iteration To perform loess smoothing Choose Statistics gt Smoothing gt Local Regression loess from the main menu T he dialog shown below appears Loess Smoothing oix m Data Options Data Frame lethanol Span fo 67 x Variable Degree of Locally fitted Polynomial y Variable fi z Family symmetric C gaussian No of Values Evaluated 50 Results Save As flast loess smooth I Plot Loess Curve ToC cance
522. ted folder pasting a folder into the page or into the selected folder deleting the Browser Page and access to its property dialog In the Object Browser S PLUS data objects are grouped into four broad categories data frames lists matrices and vectors Each category is represented by a root node object with the name data frame list matrix and vector respectively All data objects displayed in the O bject Browser will be found under these root nodes For example an object of data class design is derived from a data frame object so it would be found under the data frame root node All S PLUs functions displayed in the O bject Browser are listed under the function root node T he interface objects of S PLUS will have a root node with the same class as the object that is labeled with the class name For example menu items are found under the M enultem root node and the search path is found under the SearchPath node The objects located at the root node are the class default objects T hese objects determine the default properties of objects created with the same ORGANIZATION OF OBJECTS class T he class default properties can be modified Table 6 2 S PLUS objects manipulated by the O bject Brows Parent Child class type Example Child C lasses data frame matrix columns character complex dates double factor vector integer itspar dat itspar numeric logical rtspar single GraphSheet graphs Composite
523. tems 590 Save Select this to save the M enultem object and any other M enultem it contains in the menu hierarchy to a file Save As Similar to Save but this allows you to save a copy of the M enultem object to a different filename Menultem Select this to access the Menultem page of the Meunlten object s property dialog Command Select this to access the Command page of the Meunltem object s property dialog Show Menu In S PLus Select this to cause the menu to be displayed in the main S PLUS menu bar T his choice is available only for M enultem objects having Type M enu Restore Default Menus Select this to restore the default S PLUS menus in the main menu bar For example this will undo the effect of selecting Show Menu In S PLUS This choice is available only for M enultem objects having Type M enu Save Menultem O bject as default Select this to make the M enultem object the default W hen a new M enultem object is created its property dialog will initially resemble that of the default object except in the N ame field Help Select this to open a help page describing M enultem objects After creating a menu system in the Object Browser right click on the M enultem object which you want used asthe main menu Select Show M enu In S PLUS from the context menu to display the menu system To restore the default S PLUS menus select Restore D efault M enus in the context menu for that same Menultem object Alternatively select
524. that toolbar Help Select this to display a help page on toolbars To hide or unhide a toolbar right click on the Toolbar object and select Hide or Unhide from the context menu To selectively hide or display toolbars right click outside of any open windows or toolbars and select Toolbars from the context menu A dialog like that shown in figure 22 4 Ionbhar Giaph Coste Sheet Disin Frome Empi of Ley edsnreater Dipi Daormanrd mis wrcithhelg Foe wii i F Gao Brong how Tool ipe M Langa Bmore Figure 22 4 The Toolbars dialog 574 appears Use the checkboxes to specify which toolbars will be visible To hide or unhide a toolbar button right click on the ToolbarButton object and select Hide or Unhide from the context menu To selectively hide or display the buttons in a toolbar right click the Toolbar object and select Buttons from the context menu A dialog like that shown in figure 22 5 appears Use the checkboxes to specify which buttons will be visible in the TOOLBARS AND PALETTES toolbar Figure 22 5 The Buttons dialog Manipulating Toolbars Saving and Opening Toolbars Toolbar buttons are easily copied moved and deleted through the O bject Browser To save a toolbar to an external file right click on the Toolbar object in the O bject Browser and select Save As in the context menu Enter a filename in the Save As dialog and click OK The extension STB is added to the fil
525. the changes to our graph it now lookslike this 69 CHAPTER 3 TUTORIAL Creating 3D Now we will use the data frame ethanol to create a 3D plot Graphs 1 Use the Find button on the Object Browser toolbar to find the ethanol data frame 2 In the Object Browser click on the ethanol data frame in the left pane 3 Click on the column C in the right pane of the browser then CTRL click on E and NOx 4 Click the 3D Plots button to open a palette of available 3D plot types Plote30 Figure 3 17 The Spline button on the 3D Plot Palette circled 5 Click the 3D Spline Plot button on the palette see figure 3 17 T he 70 IN DEPTH TOUR graph will appear NOx yon 4200 2400 3 800 Figure 3 18 3D Spline plot Adding Color N ow we will add color draping to the spline plot Draping 6 Click on the maximize button in the upper right corner of the Graph sheet Figure 3 19 The 32 Color D raped Surface button on the 3D Plot Palette circled 7 Click on the mesh of the surface plot to select it On the 3D Plot 71 CHAPTER 3 TUTORIAL palette click the 32 Color Draped Surface button see figure 3 19 The graph will redraw with the new format NOK go 1w 2 400 3 800 KS sam Figure3 20 Color draping can be done with up to 32 colorsusng the 3D Plot palette or up to 64 colors using the Surface Plot dialog obtained by right clicking or doubledicking on the surface Rotating 3D Plots Now
526. the down arrow button for a list of data sets X Column Y Column Z Column W Column Specify the names of the columns to be graphed If you are graphing elements of a data frame list or matrix you can specify the position number within the data set instead of the name N ote that only the references to the data will be stored if you save the Graph sheet EDITING DATA SPECIFICATIONS To graph the result of an S PLUS expression or to have the data stored with the graph use an equal sign as the first character in the field for example X Column s 1og x The expression will be evaluated the first time that it is used From that point on the computed data will be stored within the plot T he data will be recalculated only if the expression is changed You can refer to columns in separate data sets by specifying the name of the data set along with the column or number separated by a for example fuel frame Mileage Whenever a data set name is supplied along with the column name that data set overrides the current data set in the D ata Set field Subset Rows with You can graph only a portion of the rows in the specified columns by specifying an S PLUS expression which identifies the rows to use in the analysis To use all rows in the data set enter ALL in this field The expression must evaluate to a vector of logical values T RUE values are used FALSE values are dropped or a vector of indices identifying the numbe
527. the function in a script window double click on the name to select it right click to get a menu and choose Show Dialog Example The This example looks into the structure behind the Contingency Table dialog Contingency The Contingency Table dialog in S PLUS figure22 10 is found under A Statistics p Data Summaries p Crosstabulations Table Dialog 582 DIALOGS Model Opiore Dem Fonrele ets Frome ject Petar g Sbst Rowa Aaspomeftom f Trstieaingwth fot Ferma oOo OO O Fesuls Bava As hi F Prev Feels Figure22 10 The Contingency Table dialog It has two tabbed pages named M odel and Options On the M odel page are three group boxes named D ata Formula and Results The FunctionIlnfo object for this dialog is called menuCrosstabs its property dialog is shown in figure 22 11 and is described below function Inhuman 77 e p Burcin EA horp Choad Corpa ote ee Lit EL eate deaig O a aia ect Pooler Poa Bec Bpmarilat Phi epia E i g aaa M Depi Ea e Her Figure22 11 The property dialog for the Function nfo object menuCrosstabs Function Name N otice that this value is also menuCrosstabs the S PLUS function associated with this dialog has the same name as the Functi onl nfo object To look at the code behind the function menuCrosstabs type menuCrosstabs Or page menuCrosstabs at the prompt in the Commands window Dialog Header This is the header which appears at
528. the location of the application s EXE file in the Send Graph toolbar button or menu item dialog You can make copies of this button and use it as a template to specify additional applications To specify the target application 1 Right click on the Send Graph button 2 Choose Property Dialog from the shortcut menu 3 On the Command page specify the complete path name to the target application s EXE file Use the Browse button to locate the file To use the Send Graph button 1 With your Graph sheet in focus click on the Send Graph button The contents of the Graph sheet are copied to the clipboard the target application is launched and a message appears indicating the Graph sheet is ready for pasting 2 Paste the Graph sheet in the desired location If the target application is already open S PLUS will copy the Graph sheet to the clipboard and switch into the application automatically 169 CHAPTER 7 EXCHANGING OBJECTS WITH OTHER APPLICATIONS CREATING A POWERPOINT PRESENTATION 170 You can automatically create a M icrosoft PowerPoint presentation using your S PLUS Graph sheets that have been saved to disk This option is enabled if you choose to install the PowerPoint Link option during setup T his feature only works for PowerPoint 7 0 and highe To create a PowerPoint presentation 1 3 Click the PowerPoint Presentation button on the standard toolbar or select C reate PowerPoint Presentation from th
529. the operating system e RUN Thefile specified on the Command page is opened and run as a script by S PLUS when this item is selected 587 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE 588 Manukam Ohjeet 1 REE bia nutem Commend Command Options Ben Deere os P Damat Figure22 16 The property dialog for a M enultem object Command page Menultem Text Enter the text which will represent the iten in the menu system T his does not apply to Separator items StatusBar Text Enter the text which will appear in the status bar when the item has the focus in the menu Hide Check this to make the item invisible When the item is hidden its icon in the O bject Browser appears grayed out This can also be specified through the context menu Deletable Check this to allow the item to be deleted The rest of the fields are found on the Command page seen in figure 22 16 Built In Operation Select from the list of actions associated to the built in S PLUS menus and toolbars It is not possible to easily modify this list T his field is enabled when the Action is set to BUILTIN Command Enter thename of an S PLUS function or a path and filename This field is enabled when Action is set to FUNCTION OPEN PRINT or RUN on the M enultem page Use the O bject Browser to identify the folder Parameters Edit the text in this field to specify the arguments for the function which will execute when the item is selected The easi
530. the order specified on the Pattern Color and Fill Color pages in this dialog If no rotating is specified the first pattern style pattern color and fill color specified will be used for the default Standard Bars Auto Change Choose whether to have the pattern style pattern color and or fill color change for each bar in anewly created standard bar chart Pattern styles colors and fill colors will rotate in the order specified on the Pattern Color and Fill Color pages in this dialog If no rotating is specified the first pattern style pattern color and fill color specified will be used for the default Grouped Bars Auto Change Choose whether to have the pattern style pattern color and or fill color change for each bar in each group in anewly created grouped bar charts Pattern styles colors and fill colors will rotate in the order specified on the Pattern Color and Fill C olor pages in this dialog If no rotating is specified the first pattern style pattern color and fill color specified will be used for the default The Lines page has ten color and ten style fields Enter the colors and styles in the order in which you want them to cycle The Symbols page has ten color and ten style fields Enter the colors and styles in the order in which you want them to cycle The Pattern Color page has ten pattern style fields and ten pattern color fields Enter the pattern styles and colors in the order in which you want them to cycle T
531. the relationship Conditioning on the Equivalence Ratio 5 Select E on the right pane of the O bject Browser and drag it to the top of your graph As your cursor passes over the top of the graph a rectangular drop target will appear Release the mouse button when within the rectangle The graph will be redrawn in panels representing different levels of E see figure 9 5 6 Right click on an empty space within the graph to get a short cut menu for the graph Select Multipand Under Continuous Conditioning set the of Panels to 9 and the Fraction of Shared Points to 0 25 Under Layout set the of Rows to 2 TRELLIS EXAMPLES 7 Click on the Position Size tab of the dialog Set the Aspect Ratio to 2 5 Click OK The Trellis graph now has nine panels in two rows and each panel has an aspect ratio of 2 5 The range of E used for each panel is determined using the default equal count method with a 25 overlap The algorithm picks interval endpoints that are values of the data the left endpoint of the lowest interval is the minimum of the data and the right endpoint of the highest interval is the maximum of the data The endpoints are chosen to make the counts of points in the intervals as nearly equal as possible and the fractions of points shared by successive intervals as close to the target fraction as possible Overlapping ranges typically provide greater sensitivity in detecting nonhomogeneity than non overlapping ranges The
532. the returned tree that is the desired number of terminal nodes The best tree of that size in the cost complexity sequence will be returned Pruning Method Select either deviance or misclass to determine the measure of node 4711 CHAPTER 18 FITTING STATISTICAL MODELS 478 Shrink heterogeneity used to guide the pruning For regression trees this is ignored See the on line help forshri nk tree for more details on this operation and its parameters Optimal Recursive Shrinking Check this to specify shrinking the fitted tree Shrinkage Parameter Enter a vector of numbers between 0 and 1 A sequence of shrunken trees will be determined by optimal shrinking for each value in the vector By default the vector 1 20 2 19 3 18 10 11 is used This vector is expressed in the S PLUS command syntax as 1 10 20 11 Plot Result Check this to generate a plot of the resulting pruned or shrunken tree or tree sequence New Data Optional Enter or select a data frame at which the optimal pruned or shrunken tree is evaluated This new data will be used in the pruning and shrinking operations Save As Enter thename for the object in which to save the results of the analysis If an object with this name already exists its contents will be overwritten This must be a valid S PLUS object name any combination of alphanumeric characters that starts with an alpha character is allowed The only non alphanumeric cha
533. the speed of the spinning Scale Use the horizontal scroll bar marked Scale to adjust the overall size of the data cloud Choosing Variables U se the fields labeled x y and z Select from the list box in each field to identify the variable to use for each axis WORKING WITH TRELLIS GRAPHICS Creating Trellis Graphs 209 Plots on Trellis Graphs 210 Formatting Plots within Panels 212 Formatting Plots within Panels 212 Formatting Panel Strips 214 Trellis Examples 216 References 226 Trellis graphs allow you to view relationships between different variables in your data set through conditioning Suppose you have a data set based on multiple variables and you want to see how plots of two variables change with variations in a third conditioning variable By using Trellis graphics you can view your data in a series of panels where each panel contains a subset of the original data divided into intervals of the conditioning variable For example a data set contains information on the high school graduation rate per year and average income per household per year for all 50 states You could plot income against graduation for different regions of the USA for instance South North East and West to determine if the relationship varies geographically In this case the regions of the USA would be the 207 CHAPTER 9 WORKING WITH TRELLIS GRAPHICS conditioning variable 9 98 99 09 09 09 09 09 09 09 09 09 09 Eaa n
534. thing and Curve Fitting Hide Choose whether to hide or display the graph area You can specify standard fill attributes fill color fill pattern and pattern color and standard line attributes style color and weight for the graph area This page of the Graph Properties dialog displays the name of the data window the selected columns the hide option and the scale to options for the selected plot Within this dialog you can edit the data specification options for individual plots or for all plots on the graph at once For details see section Preparing D ata for Graphing page 194 For more information on editing multiple plots at once see section Editing D ata Specifications page 187 You can specify standard fill attributes fill color fill pattern and pattern color and standard line attributes style color and weight for the panel area on the Fill Border page of the Graph Properties dialog See the chapter on Trellis graphs for details of multipanel conditioning Suppose you have a data set based on multiple variables and you are interested in comparing data series It is often convenient to view these data series in a series of panels keeping the x axis and or y axis ranges constant This is done by using the By Plot option on the Multipanel page of a graph FORMATTING A GRAPH For example suppose you have data on GNP the unemployment rate and the inflation rate over time To graph these series in a sing
535. ties of the Graph sheet Apply Style Color or Black and White applies the style settings specified under O ptions Graph Styles to your graph Align Align all of the selected objects to the first selection Distribute Place three or more objects equal distances apart on the Graph sheet Snap to Grid M ove select objects to the nearest horizontal or vertical grid M odify the data specifications for the plots on the graph Resize and reposition the graphs or your Graph sheet Exchange the sizes and position of the two selected graphs Bring the selected object in front of other object on the graph Send the selected object behind the other objects on the graph Bring the selected object forward by one level Send the selected object back by one level 285 CHAPTER 11 WORKING WITH GRAPH OBJECTS Figure 11 2 TheView menu when a Graph sheet is active 286 SUMMARY OF FORMAT AND VIEW MENUS Table 11 2 TheView menu options Option Purpose D raft Sets the drawing mode on the screen to draft to increase display speed Only affects the screen drawing and only speeds up certain line drawing Status Bar Turn on the status bar at the bottom of the Graph sheet window Toolbars Allows you to select toolbars to be visible Full Screen Displays your Graph sheet using the full screen without menu bar title bar or toolbars Zoom Enlarge a section of your Graph sheet to focus on a particular area Auto Plot Redraw R
536. tions below on redirecting text output on how to set this option The Report window supports most basic editing features such as cut copy paste and so on Character formatting is also supported in the Report window The Report window supports the following operations e Typing from the keyboard point and click highlight drag and drop etc Undo Cut Copy Paste Find Replace and so on are supported via the Edit menu and context sensitive menu right click Pasting from the clipboard including graphics and other OLE objects are supported for RTF files only In particular if you attempt to save the Report window as a text file and have graphical images pasted in it the graphics will not be saved and you will not get an error message Fonts are supported via the context sensitive menu right click and the Format menu Fonts are supported in RTF mode only User input via the keyboard may go through the Report window trickling input The Report window may be saved in one of two file formats Plain text format txt and Rich Text Format rtf Plain text can be used with the widest variety of programs and is relatively fast RT F has more capability but RTF files are bigger and slower than their plain text counterparts How to save the Report window 1 From the File M enu While the Report window is in focus Choose Save or Save As 563 CHAPTER 21 WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS 2 From th
537. tistics menu but can be launched through the Commands window or the Script window Using the customization features of S PLUS however any function may be S PLUS WINDOWS executed from a dialog that is invoked by a menu item or toolbar button 1 Randoms coml catalyst ethanol eure Pio Erang rari ae Ey peas a ee Le sr gee E Figure 2 8 TheCommands window will befamiliar to users of earlier versonsof S PLUS Script Window Tne Script window is designed for creating modifying and running scripts of S PLUS commands Like data sets and Graph sheets scripts can be created edited opened saved and printed You can type commands directly use commands generated in the History Log or copy graph objects into the Script window to create script files Script files can be used to create toolbar buttons menus and dialogs and to automate repetitive tasks Each script window has two panes T he upper pane is the program pane in which you can enter or copy commands T he lower pane is for script output When you run your script all output such as Print commands and warnings and errors appears in this pane You cannot enter information into this pane 43 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE but you can copy output into the clipboard Fid falliches Ising pl Time lous pi Timneh Randone follicles isin 2 a pi a Time eos 2 pi a Timet Guster Mar
538. tles comments and other text in agraph will resize when you resize the graph T he default is to not resize text Resize Symbols with Graph Choose whether symbols are resized when you resize the graph Set both of the above options to have text and symbols in your graph 601 CHAPTER 23 CUSTOMIZNG YOUR S PLUS SESSION automatically resized when you resize the graph Tip Brush and Spin Page Specifying Graph Styles Options Page 602 Shapes such as the open rectangle filled rectangle oval etc will always resize regardless of the setting of the Resize Symbols with Graph option Auto Add Pages C hoose whether to have pages automatically added when a series of plots are created within an S PLUS function by default This can be overridden in the graph sheet property dialog Create Editable Graphics Choose whether to have plots created within S PLUS functions translated into editable graphical objects when placed in a graph sheet When this option is on creation of graphs can be very slow If this option is off acomposite graph object will be placed in the graph sheet It can be converted to editable graphical objects at any time by right clicking and selecting from the short cut menu Specify the font font size background color and foreground color used in the Brush and Spin window To specify graph styles selection O ptions Graph Styles from the main menu Two graph styles can be defined C olo
539. to Arrange Specify the desired layout style for new graphs placed on the Graph sheet Choose N one to have all graphs set to the size and position of the default graph Choose O verlaid to have each new graph created full size on the Graph shee using the offsets specified above If you choose One Across Two Across T hree Across or Four Across the first graph is created at full size W hen you add additional graphs all graphs including the first one will be resized and positioned automatically into the selected layout style for example Two Across For more information on Arranging Graphs see the section Arranging Graphs on the Graph Sheet page 192 229 CHAPTER 10 FORMATTING A GRAPH Options Page 230 In the Graph Sheet dialog the O ptions page has the following options Mode M any commands in the S PLUS language draw plots using primitives such as lines and points T hese plots can be put into a Graph sheet in two different ways If the mode is set to Fast a single composite object will be created It will draw quickly on the screen but its contents cannot be directly edited Alternatively if the mode is set to Editable a series of editable graphical objects will be created such as lines axes and plots If the mode is set to Auto the mode set from the O ptions Graphs dialog which also can be changed via the command line toolbar will be used Composites objects can be converted to editable objects
540. to Mida Deane P Add Tot Labats r Undormby Sire Lataki Branch Size Proportional to Node Deviance Check this to size branches in the plotted tree roughly proportionally to the deviance of the node Uniformly Sized Check this to plot all branches with uniform size TREE MODELS Branch Text Add Text Labels Check this to create and add text labels to the tree plot Labels Select the type of label to be used Choose from Response Value N ode Size and Node D eviance See the on line help fortext tree for more details Tine Aigmasion Motel Regut Piot Phere Strins Predict Prag ahiak F Goat Comp lasity Pruning P Optima Fawna Shining E Fiu Rami Now Desks Ciptions Shae Ai Prune Shrink Use the Prune Shrink page to manage either pruning or shrinking of trees It page is not possible to specify both pruning and shrinking at once By default neither pruning nor shrinking is applied unless explicitly requested Prune Seethe on line help for prune tree for more details on this operation and the parameters mentioned here Cost Complexity Pruning Check this to specify pruning of the fitted tree Cost Complexity Parameter Enter a scalar for a specific subtree of the fitted object or a vector to return a sequence of subtrees minimizing the cost complexity measure so represented T he default is to compute this parameter algorithmically Size of Returned Tree Optional Enter an integer specifying the desired size of
541. to a name that starts with last For example last lm is the most recent linear regression model fit Print Results Check this to print the results of the onesample t test in the designated output window Related S PLus language functions t test 381 CHAPTER 17 COMPARING SAMPLES ONE SAMPLE WILCOXON TEST 382 Data Hypotheses Options Results T his dialog performs a one sample W ilcoxon Signed Rank test To perform a one sample Wilcoxon test Choose Statistics gt Compare Samples gt One Sample gt Wilcoxon Signed Rank Test from the main menu T he dialog shown below appears it Ine eample Silom Test Dni Opin Deis Femi F ee EnaaDiihhuran a earn Lie F Conii Gonmei Hypotheses Femi Biull Hepat Mean p ee Isnt eter Alenarte Hippi im mied E Prt Faaite Data Frame Select a data frame Variable Select the column to which the one sample W ilcoxon test will be applied Null Hypoth Mean Enter the assumed population mean of Variable Alternative Hypoth Select the alternative hypothesis For example to perform a one sided test against the alternative hypothesis that the mean of Variable is greater than the null hypothesis mean select greater Use Exact Distribution Check this to use the exact distribution of the test statistic to compute the p value Continuity Correction Check this to use a continuity correction in the normal approximation to the distribution
542. toff Loading Value Enter anumber giving the cutoff for printing the loadings Elements of the loadings matrix whose absolute value is smaller than the cutoff value will PRINCIPAL COMPONENTS ANALYSIS appear as blanks T his field is only enabled when Loadings is checked Principal Lompimre ris Annahme mE Ei Moda Faauta Pint Pranic Pin Bpi pnna F Soseplot Ei pigi tich Scii F Lowdings pele F Epi Plot Page Plots Screeplot Check this to produce a barplot of eigenvalues for each principal component Loadings Check this to produce a barplot of the component loadings Biplot Check this to produce a biplot between two components of the fitted modal Biplot Options Biplot Which Scores Enter the two components to be plotted in the form c factor1 factor2 By default a biplot of the first two components is created T his field is enabled only when Biplot is checked 503 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS Princapel L ampimerig Anaye Moda Faauta Pint Pirs Diets i Eman hi l Pradactona Predict Page New Data Enter the name of a matrix or data frame containing columns with the same names as those used in the original analysis Save Save In Enter the name of an S PLUS data frame in which the predictions of the analysis are saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that alre
543. tomobile exhaust An experiment was done on a one cylinder engine fueled by ethanol Two engine factors were studied the equivalence ratio E a measure of the richness of the air and fuel mixture and the compression ratio to which the engine is set C There were 88 runs of the experiment Here we will examine the relationship between the oxides of nitrogen and the compression ratio Creating a Loess plot and adjusting the smoothing parameter 1 Load the Example O bject Browser or use the Find button on your O bject Browser to put the ethanol data frame in a folder 2 Select the ethanol data frame in the left pane of the O bject Browser In the right pane select C then CTRL click to select NO x Open the 2D plot palette and make sure that conditioning mode is off Click on the Loess plot button 3 Make sure that all windows other than the new Graph sheet and the Object Browser are closed or minimized Then choose W indow Tile Vertical from the main menu 4 Right click on any symbol or the plotted loess line to get the short cut menu for the loess plot Select Smooth Sort Change the Span parameter to 1 Click OK An increase in the span parameter will result in a smoother line There appears to be little dependence of the oxides of nitrogen on the compression ratio H owever hidden from this scatter plot is the fact that E is varying as we move from point to point The next step is to condition the plot on E to further examine
544. tool to draw lines at an angle A radial line rotates around a fixed center point acting as a radius for an invisible circle The starting point will act as the center point Error Bars Use the Error Bar tool to draw and size error bars on your graph Error bars are initially drawn so that the bars expand symmetrically from all sides After they have been drawn on the graph you can select parts of the error bars and manipulate them individually Reference Lines You can use a reference line to highlight a particular x or y value A reference lines position is set in axes units To add a reference line you can drag it from the annotation palette ADDING LINES SHAPES AND SYMBOLS Shapes Use the Shape drawing tools to draw rectangles rounded rectangles or ellipses of any size or proportion A shape can be clear or opaque Symbols Use any of the Symbol drawing tools to place extra symbols on your Graph T hese are the same symbols used to represent data points on plots The 28 symbol types appear in blue on the palette Symbols unlike Shapes always retain their original proportions when resized 279 CHAPTER 10 FORMATTING A GRAPH 280 WORKING WITH GRAPH OBJECTS Graphic O bjects 281 Selecting O bjects 282 Summary of Format and View Menus 284 Moving and Copying O bjects 288 Using Drag and D rop 288 Using Cut Copy and Paste O ptions 288 Sizing O bjects 289 Editing O bjects 290 Editing a Single O bject 290 Edit
545. ts a gridding algorithm is used to grid the data before plotting the surface It is possible to create a wide variety of specialized plots by using the user defined function property in line plot objects If a user function is specified on the Smooth Sort page of the line plot the results computed by the function will be used when drawing the line instead of the original x and y data All of the properties of the plot related to line drawing will be used when drawing the transformed data For example the function crosshairs computes the mean and 95 confidence intervals of the x and y variables and places the computationsin a list We can use this to draw crosshairs showing the means and 95 confidence intervals for the Weight and M ileage of automobiles TRELLIS EXAMPLES Create a scatter plot with crosshairs 1 Select the fuel frame data frame in the left pane of the Object Browser In the right pane select Weight then cTRL click to select Mileage Open the 2D plot palette and click on the scatter plot button A scatter plot will appear 2 Right click on any symbol to get the short cut menu Select Smooth Sort Specify the Smoothing Type as User Type crosshairs for the function name 3 Click on the Line tab of the dialog Set the Line Style to solid Click OK Create a panel for each automobile type In a conditioned Trellis graph the computations will be repeated for the data in each panel 4 Press CTRL shift V to t
546. ts Save In Enter the name of an S PLUS data frame in which a part such as fitted values and residuals of the analysis is saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of rows as the number of observations used in the model fit then the saved values are appended to this data frame T his allows you to keep fitted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used 524 PARAMETRIC SURVIVAL Fitted Values Check this to save the fitted values from the model in the object specified in Save In Deviance Residuals Check this to save the deviance residuals The sum of squares of these add up to the deviance Pearson Residuals Check this to save the pearson residuals T hese are standardized residuals on the scale of the response Working Residuals Check this to save the residuals from the final IRLS fit Matrix Residuals Check this to save the matrix residuals Related S PLus language functions for Parametric Survival survreg print survreg predict survreg Summary survreg residuals survreg survreg control surv
547. turn the sub objects can be expanded or collapsed To open the Object Browser Figure 6 1 The Object Browser is opened by default at the tart of a sesion but if it has been closed it can be opened with the Browse button circled on the Standard toolbar Expanding a node e Click the plus symbol to the left of the node Collapsing a node e Click the minus symbol to the left of the node 142 FILTERING OBJECTS AND DATABASES m Dbyert Brorem form By i E ei The right pane displays objects in a linear list When an object is selected in the left pane the right pane lists its immediate children and displays their object attributes in a column fashion You can modify the right pane to display specific object attributes of interest such as the object data class dimension or the date it was last modified Only single selection is permissible in the left pane but multiple selection is supported in the right pane 3 LJ aihe H ba a s JH bs T Figure 6 2 Thelig last gim is sected in the left pane and the right pane hows the contents of thelig Filtering Objects and Databases The number of objects in S PLUS is immense and many are not of direct interest The O bject Browser allows you to filter a selected set of the objects you wish to see As an example data objects and functions in S PLUS are stored in various databases and generally you will only be interested in viewing your own data and funct
548. tyles dialogs T he colors are specified as a range or as a list of special colors Vary Color By Choose from None x y z w or Other Column Colors To Use Choose from Range or Special used only if the third column has numeric data Minimum Color Minimum color in the shading range Maximum Color M aximum color in the shading range Column Choosea column from thelist D ata Set C hoose a data set from the list Special C olors Specify custom colors You can also show an extra dimension of data by varying symbol styles T his is appropriate when the vary symbol styles column is a factor As with vary color the symbol style will be determined by mapping the factor level to the varying symbol styles on the Symbols page of the Options Graph Styles dialogs Vary Style By Choose from None x y z w or Other Column Column Choose a column from the list D ata Set C hoose a data set from the list On the Smooth Sort page of the plot property dialog you have the following smoothing options Pre Sort D ata Choose from None X Y on X X Y on Y X only and Y only If a sorting option is selected data is sorted before any smoothing or rendering of the line is done Smoothing Type Choose from None Least Squares Robust Kernel Loess Spline Friedman s Super None uses no smoothing Least squares uses the m function to perform a least squares linear fit 321 CHAPTER 12 EDITING PLOT PROPERTIES For Loess Fr
549. uare root of the absolute values of the 417 CHAPTER 18 FITTING STATISTICAL MODELS 418 Options Partial Residual Plots residuals versus the fitted values This plot is useful for checking for the constant variance assumption of the model Response vs Fit Check this to display a plot of the response variable versus the fitted values Theline y x is also drawn on the graph Residuals Normal QQ Check this to display aN ormal quantile quantile plot of the residuals Residual Fit Spread Check this to display a residual fit spread plot T his is a visual analog of the multiple R squared statistic It compares the spread of the fitted values to the spread of the residuals Cook s Distance Check this to display a plot of Cook s distance values versus the observation number Include Smooth Check this to display a smooth curve computed with loess smooth on the Residuals vs Fit Sqrt Abs Residuals vs Fit and Response vs Fit plots See the on line help for oess smooth for details Include Rugplot Check this to display a rugplot on the Residuals vs Fit Sqrt Abs Residuals vs Fit and Response vs Fit plots A rugplot is a sequence of vertical bars along the x axis that mark the observed x values Number of Extreme Points to Identify Enter the number of extreme points that will be identified on the Residuals vs Fit Sqrt Abs Residuals vs Fit Residuals Normal QQ and Cook s Distance plots The row names from the
550. ula Confidence Interval Type Select a confidence interval type e log for intervals based on the cumulative hazard or log survival T his is the default e plain for standard intervals log log for intervals based on the log hazard or log log survival NONPARAMETRIC SURVIVAL Lower Limits Select a specification for the modified lower limits to the curves usual peto or modified T he upper limits remain unchanged Meenneuartii Gerri All Model Optors Pesis me Prrd Perih Summar Opinna F hoitui M incide Cerone Times P Long un Hew Tite l Doang Fado Results Page Printed Results Short Output Check this to print a short summary of the model results in the designated output window T hisincludes the number of observations number of events the mean survival and its standard error and the median survival with confidence limits for the median Long Output Check this to print a long summary of the model results in the designated output window This produces tabled output including columns for the Survival estimates the standard errors of the estimates and confidence bounds for the estimates Summary Options Include Censoring Times Check this to include censoring times in the output T his field is ignored if a vector is specified in N ew Times New Times Enter a vector of times listed in increasing order and having no missing 509 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES
551. ula is displayed in an edit field which spans two columns of the dialog instead of one column 581 CHAPTER 22 CUSTOMIZNG THE USER INTERFACE Modifying Property objects and Functioninfo objects may be modified through the Dialogs same dialogs which are used to create them To modify a Property object open the O bject Browser to a page with filtering set to Property Right click on the Property object s icon in the right pane and choose Properties from the context menu Refer to the previous sections for details on using the property dialog To modify aFuncti onl nfo object open the O bject Browser to a page with filtering set to Functi onl nfo Right click on the Functioninfo object s icon in the right pane and choose Properties from the context menu R efer to the previous sections for details on using the dialog Displaying T here are several ways to display a dialog in S PLUS Dialo gs Locate the associated function in the O bject Browser and double click on itsicon If afunction is not associated with aFuncti onl nfo object then double clicking on its icon will cause a default dialog to be displayed Click on a toolbar button which is linked to the associated function Select a menu item which is linked to the associated function T hisis described in the section M enus Use the function guiDisplayDialog in a Script or Commands window T his is described in the S PLUS Programme s Guide Write the name of
552. urrent Help Figure2 11 An S PLUS dialog in this cas for performing multiple comparisons 47 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE S PLus PALETTES AND TOOLBARS T here are ten main toolbars built into S PLUS Toolbars contain buttons that are shortcuts to menu selections You can use toolbar buttons to perform standard file operations such as opening a file or saving a file or operations highly specific to S PLUS such as making immediate changes to selected plot properties such as text font and color changes Every toolbar can be dragged into an open window as a floating palette When you pause the mouse over a toolbar button ToolTips will appear indicating the function of the button T he following is a complete gallery of the ten toolbars Standard Toolbar Table 2 3 The Standard toolbar for the U ndo Redo Commands and H istory log options see chapter 21 Working with Script Commands and Report windows For details on the Restore data object and N ew data frame buttons seechapter 4 Using Data Windows See the graphics section starting with chapter 8 Creating a Graph for full descriptions of the graphical options Standard New Open Save Cut Copy Paste Undo Undo list Redo N ew Data New Display the Frame Browser History log 2D Plot 3D Plot Condition palette palette ing mode Draft mode PowerPoint presentation 48 Print Redo list Restore data object Commands Commands win
553. ut window T his includes the model call a five number summary of the deviance residuals the null and residual deviance values along with their degrees of freedom and the degrees of freedom for each term in the modd ANOVA Table Check this to display an analysis of variance table T he sums of squares in the table are for the terms added sequentially Type sums of squares Correlation Matrix of Estimates Check this to display the correlation matrix of the regression coefficients This is available only if Long O utput is selected Save In Enter the name of an S PLUS data frame in which fitted values and residuals are to be saved If an object with the name you enter does not already exist in database 1 then it will be created If you enter the name of a data frame that already exists in database 1 and this data frame has the same number of GENERALIZED LINEAR MODELS rows as the number of observations used in the model fit then the saved values are appended to this data frame T his allows you to keep fitted values from a model with the original data or to keep the residuals from a number of different models for the same data in one data frame If you give the name of an existing S PLUS object that is not a data frame or is not the appropriate size then a warning is issued and a modified name is used Fitted Values Check this to save the fitted values from the model in the object specified in Save In Working Residuals Che
554. utside the Graph sheet to deactivate it T he embedded G raph sheet is displayed in your document To edit the embedded Graph sheet in place 1 In your client application document double click the embedded S PLUS Graph sheet or select the Graph sheet choose O bject from the Edit menu then choose Edit The embedded Graph sheet remains in the client application but the menus and toolbars change to those used by S PLUS Edit the Graph sheet using the S PLUS menus and toolbars When finished click anywhere outside the embedded object to return to the client application s menus and toolbars Updating You can update changes to an embedded S PLUS Graph sheet without leaving Embedded the current S PLUS session when you have opened an embedded S PLUS graphsheet Graphs 168 EMBEDDING S PLUS GRAPHICS WITHIN OTHER APPLICATIONS To update the embedded Graph sheet w Click the Update button this button replaces the Save button when editing an embedded Graph sheet in S PLUS The Update button updates the embedded graph in the client application document where it is enbedded or w Select the Save Copy As option on the File menu Save the embedded graph to anew Graph sheet file SGR file on disk Copying Graphics S PLus has a Send Graph button that copies the current Graph sheet to the Using the Send clipboard then switches automatically into a specified application eg Graph Button Word for pasting You just specify
555. utton a description appears at the bottom of the screen in the status bar Like menu options some toolbar buttons may not be available at all times Inactive toolbar buttons are dimmed for example the Undo button is dimmed if there is nothing to undo Several buttons display a palette of options when selected For example the 2D Plots button will display a palette of 2D plot types Tool palettes remain in view until you click the toolbar button again or click the palette s Close box Leaving a tool palette in view is convenient when you are experimenting with different options 39 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE S PLus WINDOWS Object Browser TheS PLUS user interface contains six types of windows the O bject Browser D ata window Graph sheet Commands window Script window and Report window T hese windows allow you to easily organize your work session work with data scripts and graphs simultaneously and automate repetitive tasks The Object Browser provides a detailed map of S PLUS This two paned window displays the data sets graphs functions and other objects in your S PLUS session T he left pane shows a hierarchical tree view of the objects in the current session Branches of the tree may be expanded and collapsed to show any level of detail The right pane displays the objects which are contained in a left pane selection just as in the Windows 95 Explorer By using the O bject Browser you can
556. veral arrows and then a line plot the dialog will contain arrow properties H owever if you change the arrow line color here the line color used for the plot will also be affected because line plots and arrows have the Line Color property in common ChooseOK 291 CHAPTER 11 WORKING WITH GRAPH OBJECTS ARRANGING OBJECTS ON THE GRAPH Overlapping Objects Aligning Objects 292 If objects overlap some objects may be completely or partially covered by others You can change the order of overlapped objects by bringing certain objects to the front or sending others to the back To use Bring to Front 1 Select the object or objects you wish to bring to front or send to the back Bring to Front Send to Back 2 Click on the Bring to Front or Send to Back buttons on the toolbar or from the Format menu choose Bring to Front or Send to Back To bring objects forward or backward by single increments one level at atime choose the Bring Forward or Send Backward options from the Format menu You can align selected graph objects automatically using S PLUS Align and Snap to Grid options Snap to Grid lets you align your objects to a grid S PLUS s grid is made up of a series of invisible horizontal and vertical gridlines By default there are 12 invisible gridlines per inch and 5 per centimeter you can change the default under O ptions Graph O ptions As you move an object it will snap to the closest intersection of the in
557. view all data sets currently in windows by selecting M ore Windows from the Window menu All windows currently open in S PLUS will be listed Click on any window to make it the active window You can move the active window by dragging the title bar You can resize the active window by dragging its frame M aximized or minimized windows must be restored to their actual size before they can be moved or resized To restore a window to its original size and position click the window s restore button S PLUS also gives you control over column widths and row height column names and descriptions and row numbers You can specify data type CREATING A NEW DATA WINDOW and numeric formats and precision Table 4 1 T he D ataFrame toolbar Align left Align center Align right Increase D ecrease Change data precision precision type Insert Column type column to insert Remove Clear Remove row Clear row column column Sort Sort Adjust Increase D ecrease ascending descending column column column width to fit width by one width by one Creating a New To create a new Data window Data Window 1 From the D ata menu choose N ew D ata O bject 2 TheN ew Data O bject dialog box will appear allowing you to select D ata Frame M atrix or Vector or w To create a new data frame click on the N ew Data Frame button on the Standard toolbar to the left of the O bject Browser button Data Objects S PLus has four basic types of
558. visible horizontal and vertical gridlines To align objects using Align 1 Select the anchor object This object s position will be used to align other objects 2 SHIFT click or CTRL click to select the objects you wish to align 3 From the Format menu choose Align 4 Choose the desired alignment from the Align menu ARRANGING OBJECTS ON THE GRAPH Aligning Objects Using Snap to Grid Distributing Objects Saving Defaults You can use Align to align your graphs on the Graph sheet Alternatively you can use S PLUS automatic graph layout feature Arrange Graphs to position multiple graphs on the Graph sheet To align objects using Snap to Grid 1 From the Format menu choose Snap to Grid 2 Select the object you wish to move 3 Drag the object slowly to the desired position The object will snap to the nearest horizontal and vertical gridline automatically You can use the Distribute option to place three or more objects equal distances apart on the Graph sheet The two outermost selected objects determine the size of the areain which to distribute the selected objects T his option is useful when you have multiple objects for example comments that you want to have evenly spaced on the Graph sheet You can use Align to line up your objects then use Distribute to space them evenly To distribute objects 1 Select three or more objects to be distributed 2 From the Format menu choose Distribute 3 From the Di
559. w Chapter 22 Customizing The User Interface Overview Toolbars and Palettes Creating T oolbars Creating and M odifying Buttons M odifying T oolbars Displaying T oolbars M anipulating T oolbars Saving and O pening T oolbars Dialogs Creating D ialogs M odifying Dialogs Displaying Dialogs Example The Contingency T able D ialog M enus Creating M enu Items M odifying M enu Items Displaying M enu Items M anipulating M enu Items Saving and O pening M enus Example Customizing the C ontext M enu Using the Classi nfo O bject Properties of the C lassl nfo O bject Creating and M odifying a Class nfo O bject Chapter 23 Customizing Your S PLus Session O verview 552 554 555 555 556 558 558 560 560 561 563 567 567 568 568 569 572 574 575 575 576 577 582 582 582 586 586 589 590 590 591 591 595 595 596 597 597 xiii CONTENTS Changing D efaults and Settings 598 Saving O bject D efaults 598 Specifying G eneral Settings 598 Specifying Command W indow Settings 600 Specifying Undo amp History O ptions 600 Specifying T ext O utput W indow Settings 601 Specifying G raph O ptions 601 Specifying G raph Styles 602 Specifying C olor Schemes 603 Redrawing Plots Automatically 604 Chapter 24 Index 605 T rademarks 617 xiv WELCOME TO S PLus Introduction Introduction 15 Installation 16 N etwork Installation 17 System Requirements 18 Help Support and Learning Resources 19 Getting Help 19 W h
560. w 545 The Script Window 547 Working with Scripts 547 Using Find and Replace 552 H iding and Unhiding Scripts 554 Context Sensitive H elp 555 Using the Right Button M enu 555 Show D ialog 556 Time Saving Tips for Using Scripts in S plus 558 H istory Log 558 Dragging Graph O bjects into a Script W indow 560 Dragging Function O bjects into a Script Window 560 The Commands Window 561 The Report Window 563 With the Script window scripts arbitrary S PLUS commands can be written to automate the more repetitive aspects of analyzing data and creating graphs Each Script window has an output pane that displays output from the running script and a program pane that is used to typein the commands that make up the script Scripts allow you to access S PLUS s programming language to write and commands to be executed to import or export data transform the data run analyses create modify or print graphs etc You can execute scripts from within S PLUS or from another application via DDE or by calling S PLUS and passing the script name on the command line Script windows are an alternative to the Commands window The Commands window is interactive and commands typed in the window are immediately evaluated through the interpreter with the output shown below 545 CHAPTER 21 WORKING WITH SCRIPT COMMANDS AND REPORT WINDOWS 24 Script program each command T he Script window on the other hand lets you type a set of commands and funct
561. we will rotate the plot to explore the different features of our data 8 Click on the surface of the plot to select itO n the 3D plot palette click the D ata Grid Surface plot button see figure 3 21 to draw a wireframe plot 9 Click inside the plot area not on the wireframe Four circles and one triangle appear 10 T he circles allow you to rotate your graph horizontally T he triangle rotates vertically D rag a triangle or a circle and try it for yourself 11 We can also select a 4 panel rotation view of our graph Click on the wireframe plot to select it Then click on the 4 Panel Rotation button on the 3D palette see figure 3 21 T his view allows you to compare the surface plot from four different angles simultaneously 12 Additionally you can click near the surface plot in the lower left and the green circles and triangle will appear Now you can rotate this graph and watch the others rotate as well 72 IN DEPTH TOUR 13 Close the 3D graph window and the D ata window Figure 3 21 The Data Grid Surface button on the 3D Plot Palette circled and the 4 Pand Rotation button marked by an ellipse 73 CHAPTER 3 TUTORIAL Using Trellis Graphics for Multipanel Conditioning 74 Suppose you have a data set with multiple variables and you want to see how plots of two variables change with variations in one or more conditioning variables Exclusive to S PLUS Trellis graphics are designed to display
562. will automatically expand if the text starts extending beyond the box boundary Choose Date Stamp from the Insert Annotation menu S PLUS Opens an edit box for you to enter text Type the desired text Press ENTER to create a new line To end editing and save results click outside the edit box Alternatively you Can press F10 Or press CTRL ENTER To exit without saving press the 269 CHAPTER 10 FORMATTING A GRAPH ESC key Note Adding a Legend Position Size Page 270 The date stamp is identical to the text boxes you can create with the Text tool T he only difference is that the date amp time stamp optionsin the property dialog are turned on A legend is a combination of text and graphics that explain the different plots on the graph In a multipanel graph the legend only appears in one pandl To add a legend E w Click the Legend button on the Graph toolbar or choose Legend from the Insert menu If you have more than one graph on the Graph sheet select the desired graph before choosing Legend from the Insert menu To remove the legend click again on the Legend button To format the legend box w Doubleclick the legend margin outside of the legend items or select the legend and choose Selected O bject from the Format menu In the Legend dialog you can specify the legend position and formatting for the legend box Alternatively you can right click on the legend and select pages of the Le
563. window Printed results include sums of squares of the factor loadings the size of the data the names of the components in the fitted moda object and the call that created the model object Component Importance Check this to include the importance of each factor in the printed results Loadings Check this to include the loadings matrix with the printed results Cutoff Loading Value Enter anumber giving the cutoff for printing the loadings Elements of the FACTOR ANALYSIS loadings matrix whose absolute value is smaller than the cutoff value will appear as blanks T his field is only enabled when Loadings is checked Faetie Apher E Model Optus Resis Pios Predict Fin Alot Opuone F laingia r I Plots Page Plot Loadings Check this to display a barplot of the factor loadings for each factor Biplot Check this to produce a biplot between two factors of the fitted model T he biplot shows the relation of the factors to both the original variables and the original data This field is enabled only when the number of factors to be fitted is greater than one Biplot Options Biplot Which Scores Enter the two factors to be plotted in the form c factor1 factor2 By default a biplot of the first two factors is created This field is enabled only when Biplot is checked 497 CHAPTER 19 USING MULTIVARIATE SURVIVAL AND TIME SERIES MODELS Faetie Anahi Mode Opiors Fiesi Piger Data e ECOL Smeh T Fredidion
564. window used in the computations T he default is calculated using the formula log length x base 2 1 The standard error for a Gaussian window is width 4 For the other windows width is the width of the interval on which the window is non zero From and To Specify the lower bound minimum and upper bound maximum for the data you wish to use The default Auto automatic calculation uses the minimum and maximum values in the data provided Cut Specify the fraction of the window width that the x values are to be extended by The default is 0 75 for the Gaussian window and 0 5 for the other windows Remove NA s Specify whether missing values should be removed before estimation or not For histograms you can specify the standard fill attributes color pattern and background and border attributes style color and weight Draw Bars Specify whether to draw borders for the individual bars of the histogram Draw Histograms Specify whether to draw histogram D isable this to draw only the density line For density lines you can specify the standard line attributes style color and weight Draw Line Specify whether to draw the density line Disable this to draw only the histogram Specify the names or column numbers of your X and Y columns If you specify only one column of data X or Y an integer sequence will be used for the missing column of data For line and scatter plots you can specify standard line attributes style
565. wing tool T he object to be drawn appears as a small symbol on the lower right side of the mouse pointer Place this tool on the plot and click and drag to insert a drawing object of a specified size Figure10 8 Themouse pointe asa drawing tool to add a filled circle To add an object from the Annotation palette 1 Click the Annotation button on the Graph sheet toolbar A palette of drawing objects is displayed 275 CHAPTER 10 FORMATTING A GRAPH 2 Drag and drop the desired object onto the graph The object remains selected until you draw another object or click somewhere else on the sheet 3 Resize the object by selecting it and dragging on one of the corner selection knobs M ove the object by selecting and dragging in the center of the object To edit other properties of the object double click on it to open the property dialog or right click and select the relevant page of the dialog from the menu 276 ADDING LINES SHAPES AND SYMBOLS Summary of the Annotation Palette Annotation Radial Lines Tool Filled up triangle Box down triangle X Diamond Label Point Tool Arrow Tool Vert Ref Line Rounded Box Up triangle D own triangle Circle Box up triangle Filled diamond Cross Male Select Row Tool Arcs Tool H oriz Ref Line Ellipse Shape Plus Box X Up Down triangles Filled square Filled dwn triangle Ant Female Figure 10 9 The Annotation toolbar has buttons for T
566. with toolbars by the methods in this section take an occasional moment to review the program code generated in the history log See the S PLUS Programmers Guide for a systematic explanation of the program code To create a Toolbar object first open the Object Browser and filter by Toolbar to see the toolbars and toolbar buttons To create a new toolbar right click on the default object icon labeled Toolbar in the left pane of the Object Browser Select New Toolbar from the context menu Alternatively right click in the S PLUS application window outside of any open document window and choose N ew Toolbar from the context menu The New Toolbar dialog appears as shown in figure 22 1 Tesiner Hame jako Tonies fortis Fokus ae Browns Donmeni Typs Al Ligue nig tog can miske a new bugbe iom an acaring inbiat or oeae a bianik ons bim siah Cancel Figure 22 1 TheN ew Toolbar dialog 568 TOOLBARS AND PALETTES To modify the default settings that appear in this property dialog in the future right click on the default object icon choose Properties from the context menu fill out the dialog with the desired defaults and click OK Toolbar Name Enter aname for the new toolbar M ake Toolbar for this Folder Enter a path for a folder directory T he new toolbar will contain a toolbar button for each filein the indicated folder Use the Browse button if desired to identify the fol
567. xample if the active window the highlighted window contains an S PLUS D ata window the menus will display options useful for operating on Data windows Graph sheet and programming options will be absent or dimmed W hen you choose one of the main menu options alist of additional options will drop down You can choose any of the active options in the list D immed options are not available until you select the appropriate object in the appropriate window Menu options with a gt symbol at the end of the line display a submenu when selected M enu commands with an ellipsis after the command will display a dialog box when selected For more information on customizing S PLUS s menus see Chapter 22 Customizing the U ser Interface To choose a menu option PA Point to the desired menu option and click the left mouse button or lined key in the desired menu option To cancel amenu click outside the menu or press Esc When you click the right mouse button a shortcut menu for the selected object is displayed Shortcut menus contain options specific to the selected object The shortcut menu appears to the right or left of the mouse pointer To close a shortcut menu without choosing an option click outside the menu Or press ESC 35 CHAPTER 2 WORKING WITH THE GRAPHICAL USER INTERFACE Specifying Sometimes choosing a menu option or clicking the mouse displays a dialog Options in You can use dialogs to specify informat
568. xplicitly save it in PowerPoint 171 CHAPTER 7 EXCHANGING OBJECTS WITH OTHER APPLICATIONS 172 CREATING A GRAPH S PLUS graphs Creating a Graph Sheet O pening an Existing Graph Sheet Methods for Creating a Graph Viewing Graph Sheets Adding a Plot to aGraph Hiding Unhiding and D eleting a Plot Editing D ata Specifications Adding or Replacing D ata in the Plot Changing Columns and D ata Sets Placing Multiple Graphs on a Graph Sheet Adding a Graph to an Existing Graph Sheet Combining Graphs from M ultiple Graph Sheets Arranging Graphs on the Graph Sheet Preparing D ata for Graphing Projecting a 2D Plot onto a 3D Plane Combining Multiple 2D Plotsin 3D Space Combining 2D and 3D Plotson OneGraph Brush and Spin Highlighting and D ownlighting Points Brush Symbols and Size C ontrol M ore Brushing O ptions Spinning Data 174 175 175 177 182 183 186 187 187 188 190 190 191 192 194 201 201 203 204 204 205 205 206 173 CHAPTER 8 CREATING A GRAPH S PLus GRAPHS An S PLUS graph consists of one or more plots S PLUS offers a wide variety of plot types which you can combine into multiple plots on one graph or on one page You can create graphs from data in S PLUS D ata windows or from data imported from a number of applications including Excel Lotus dBase SAS SPSS and text ASCI1 files See the chapter Importing and Exporting data for detailed information on importing data I
569. y P ppm ste Property P popa m Property F SpeopPanthose Propart F spopDeseGrou Peopany Biapcp Taste Pogai I appa iiiu Peoparty Pi spp Pooparty Beppe Ver Property Pi species Ver Papert SSE Figure22 6 TheObject Browse showing all Property objects The relationship between the function the properties and the dialog is defined by aFunctionInfo object A Functi onl nfo object is a simplified dialog template but contains the map of the properties and function arguments In addition a Functionlnfo can be used to override some characteristics of the specified property objects Filter by FunctionInfo in DIALOGS w Ua Hereeere ja Contents ot Findannia the O bject Browser figure 22 7 to se objects of this type Fa i ot fi A mi a Erei Funciandaky Bi menice Fin ciana hin Fun chordie Funchordrig Funchondnig Funchordsko Fun chordie Fun chordaia Fincanda Fun chaning Funceordeis Funchordedo Figure22 7 The Object Browser showing all Function nfo objects Creating Dialogs W hile it is not hard to create or modify dialogs through the user interface as shown in this section it is easier to work with dialogs using the S PLUS programming language As you work with dialogs by the methods in this section take an occasional moment to review the program code generated in the History log M ost commands in the history log have the same argument names as the prompts of
570. y axis is left untitled D escription S PLUS uses a column description if one was specified otherwise the title is left blank 267 CHAPTER 10 FORMATTING A GRAPH Table 10 2 Automatic titling options Name S PLUS uses a column name if one was specified otherwise the field is left blank Column S PLUS uses the D ata window name followed by the column number Spec S PLUS uses the column name or number specified in the D ata to Plot page For example if you entered AGE for the x axis in the plot dialog S PLUS would print AGE for the title If you wish to use the symbol as part of your title you must type it in twice For example typing J ANUARY 12 resultsin thetitle JAN UARY 12 Adding 2D Axis in S PLus you can place axis titles on your graph Axis titles are convenient Titles because they are positioned automatically See section Adding Multi line Text page 261 for information on editing and formatting axis titles To add a 2D axis title 1 Select the axis to which you want to add a title 2 From the Insert menu select Titles 3 From the Titles submenu choose Axis S PLUS opens an edit box for you to enter text 4 Type the desired text Press ENTER to create a new line To end editing and save results click outside the edit box Alternatively you Can press F10 Or press CTRL ENTER To exit without saving press the ESC key Adding 3D 3D axes titles are different from 2D axis titles They canno
571. your data in a series of panels using conditioning options Each panel contains a subset of the original data corresponding to intervals of the conditioning variables In S PLUS most graphs can be conditioned T he data columns used for each plot and for the conditioning variable s must be of equal length T he axes specifications and panel display attributes for example fill color are identical for each panel Now we will apply multipanel conditioning to our previously created loess plot 1 Make sure the window containing the Exenvirn data is open and that the loess graph showing the relationship between radiation and ozone is open M inimize the O bject Browser Select Tile Vertical in the Windows menu The loess plot and the data should be in the main window 2 In theD ata window select both TEM PERATURE and WIND 3 Click and hold the mouse over the data in the columns not the headers until the rectangular shadow appears then drag them over and drop on the rectangular drop target the long bar which will IN DEPTH TOUR appear at the top of the graph 1 2 3 4 5 l 7 B B Figure 3 22 Highlight the TEM PERATURE and WIND columns and drag them to the rectangular drop targe at the top of the graph The Trellis graph in figure 3 23 shows the relationship between O zone and 75 CHAPTER 3 TUTORIAL Solar Radiation conditioned on both W ind Speed and Temperature The Relationship Between Radiation
572. ype such as Graph sheets 4 ChooseOK to load the file into a window 176 METHODS FOR CREATING A GRAPH METHODS FOR CREATING A GRAPH Creating a Graph Using Plot Buttons In S PLUS there are several methods for creating graphs You can select data and click a plot palette button you can drag and drop plot buttons onto graphs and then drag data onto the plot buttons or you can use the Graph option on the Insert menu The 2D and 3D Plots buttons are available on the standard toolbar for creating graphs quickly W hen you click the 2D or 3D Plots button a palette of plot buttons appears For a description of each plot move the mouse cursor over each button in the palette A text description of the plot appears in the status bar at the bottom of your screen When a new graph is created using a plot button a new Graph sheet is automatically opened in anew window To create a graph using a plot button 1 Click the 2D or 3D Plots button on the toolbar to open a palette of available plot types Figure8 1 ThePlot2D and Plot3D buttons on the Standard toolbar 2 Open aD ata window or O bject Browser containing the data to plot 3 Select the data columns you want to plot Use ctrt click to select discontiguous columns The order in which columns are selected determines their default plotting order 4 Click the desired plot button on the palette If you have added your own palette buttons using S PLUS s customizable optio
573. ypeCTRL Z or click the Undo button to undo the last cell edit Editing Time Time series objects are displayed in the same way as standard vectors Series Data matrices or data frames but contain time series information instead of row names in the row names labels column If there is a units attribute such as seconds years or milli seconds associated with the time series it will appear at the top to this time series column Ele Eq Siew jon Foma Daa gehi Graph Options Windows Pale Diela a sele e dS E a i a a l a OCORRE TE wij E a lal 1221 00 182200 1623 00 1625 00 1326 00 1607 00 1528 00 1829 00 1630 00 Figure 4 2 Sample time series data 98 ENTERING DATA To edit time series data sets 1 Doubleclick on the top of the time series column T his will bring up a dialog which allows you to edit the particular type of time series data currently in view 2 After changing the properties using this dialog and clicking OK the information in the time series column will update to reflect the changes Lear I Sas 1421 intih fi 1 End 1434 Justincation Fig Eroquancy f Cegatan Figure 4 3 Time sies properties dialog Editing individual time series elements The individual elements of regular time series those created with the ts rts or cts functions se the Programme s Guide for more details are determined by the parameters of the time series and cannot be edited dire
574. ze or underline the selected text or 1 Right click on the text to display the shortcut menu Choose Superscript and Subscript to format the text Choose Symbol to open a dialog to add or edit symbols and Greek characters C hoose Font to open a dialog to edit font type or 1 Select Font from the Format menu to open the default Font dialog 262 ADDING MULTI LINE TEXT Date Stamp Position Page Font Page For more information see section Adding Special Characters and For matting Text page 265 To align text 1 Select the anchor text T his text will be used to align other text 2 sHIFT click the text you wish to align 3 From the Format menu choose Align 4 Choose the desired alignment from the submenu To delete text 1 Select the text 2 Press the DELETE key Alternatively you can select Clear from the Edit menu or you can right click and select Cut Use D ate Select this option to add a date stamp to the text Use Time Select this option to add atime stamp to the text On the Position page the following options appear X Position and Y Position The positions of the text are determined by specifying the distance from the bottom left hand corner of the graph area to the bottom left hand corner of the text box Selecting Auto will center the text Rotation Specify the angle at which to display the text 0 360 degrees You can also rotate the text manually see To rotate text earlier in this sect
575. zed Linear Models gim plot glm predict glm summary glm Other related S PLUs language functions gam Im loess 464 GENERALIZED ADDITIVE MODELS GENERALIZED ADDITIVE MODELS T his dialog fits generalized additive models It calls the gam function and its print summary plot and predict methods Generalized additive models are an extension of generalized linear models T hey use scatterplot smoothers to nonparametrically model predictor terms allowing the data to suggest nonlinearities in the terms To fit a generalized additive model Choose Statistics gt Generalized Additive Models from the main menu The dialog shown below appears Geseliced Addes Wiodels BE tinzel Optom Pende Pior Pedis Ora Mois Daia Frame Familiy faas une Weer hhs Link f denki Subset Aina with i F P One Pees at Messing akon Saree Mace Obed Seat Mi fim 75m Famila Formals Creates Fools _Enscel Apply fennel 1e Model Page Data Data Frame Select a data frame gt Tip You can type into the D ata Frame edit box any expression which evaluates to a data frame 465 CHAPTER 18 FITTING STATISTICAL MODELS 466 Formula Model Weights Enter the column that specifies weights to be applied to all observations used in the linear regression To weigh all rows equally leave this blank Subset Rows with Enter an S PLUS expression which identifies the rows to use i

S-PLUS User's Guide

Contents

Download Pdf Manuals

Related Search

Related Contents