General introduction
Contents
1. o 72323 Ex 14 Boxplot EEE 1 25 50 80 00 xo a O 2 Extreme values LO HI UR E adjacent LO HI Hi outliers JU GE VD E Ex 15 Boxplot Part90 4 Participation avril 1990 Tot 40 5 2T Tiz O xX x Extreme values LO HI GE adjacent LO HI Oo outliers GE Hi outliers NW ZG SH Ex 16 Parallel Boxplots 23 85 69 90 RefArm x Roth X OO 2 ARM 2x x Ex 17 Parallel Boxplots 0 95 69 90 RefArm Roth x 0 3 ARM PELec x K XO PlArmP 1 x 2 Ex 18 Boxplot divison 9 1 00 4 00 Extreme values LO HI PA WY adjacent LO HI PA WY Stemleaf divison 9 Legend 110 stands for 1 00 alo for 4 00 1 000000000 0000000000000000 2 2 3 000000000000 3 410000000000000 Density line for divison 9 A density line is a kind of one line histogram showing concentrations Let us examine another density line shown together with a boxplot of the same variable Ex 19 177 countries Boxplot Urb 11 Urbanization 5 0 100 0 x a x A symbol corresponds approx to 1 0 occurrence s This is a coded density line the four symbols shown code frequencies at specific locations as the legend says the lightest symbols corresponds here to more or less one occurrence i e one country Ex 20 3113211 523336223442141624 34462111326213222221351125233322 52233 12 212 1 2 This is another form of the density line2. principle these observations should be identified and named In this case there is not enough room to do so on a single stem line therefore EDA simply informs you that there are 27 countries on that stem In the next example there is enough room to show case identifiers i e Swiss canton abreviations Ex 4 Stemleaf ALPS 1 Initiative of the Alps rail transit Legend 318 stands for 37 65 614 for 63 78 lo VS FR VD 318 41458 51122355567799 61000124 hi UR Ex 5 30 countries Stemleaf PGrow 4 Population Growth Legend 410 stands for 0 30 1010 for 1 10 4 2100 0 0 01000 2100000000 4100000 61000 8100 101000 hi ALBA TURQ AND Below you will find a stem and leaf plot as it is produced by SPSS Ex 6 AGE Age of respondent Valid cases 959 0 Missing cases 2 0 Percent missing Frequency Stem amp Leaf 2 00 l amp 98 00 2 000000011111112222222333333344444 108 00 2 55555555556666667777788888889999999 100 00 3 000000000011112222233333444444444 97 00 3 55555555666667777778888888999999 97 00 4 00000011111111222222333334444444 99 00 4 555555555566666677777888888999999 63 00 5 000011111222233333444 77 00 5 55566666777788888889999999 40 00 6 00011122233344 53 00 6 555666777888889999 35 00 7 000122233444 56 00 7 5556666777888888999 33 00 8 0001122344 1 00 Bus 6 Stem width 10 Each leaf 3 case s 2 The parentheses and the star are used to signal th
3. showing the same information using single digits for every location i e a 3 means 3 countries A star is shown if more than 9 observations are found at the same location Ex 21 183 countries Trace of Urb 5 Urbanization Range 5 00 100 00 Groups Continents g Asia Africa Europe N amp C Am S Am AusOcea x ES x x x x O x q x Xx x O x O x x x x N 39 53 30 31 15 15 EDA 1 6 EDA Software First steps Before starting to work with the EDA package you need to know how to call EDA on your computer and how to write EDA commands How to write EDA commands You interact with EDA using simple commands There is no difference between commands written in lower or upper case letters In the various examples and in the manual however we will always use upper case letters for commands and options Lower case letters will be used for parts of commands you should supply variable names etc For clarity all command line examples will be preceded by the gt symbol This symbol is not part of the command and should never be typed For instance gt GET name GET is the name of the command to be typed in upper or lower case letters name you should supply a valid name name of a work area data set gt GET SET2 is an command as you might type it i e SET2 is a work area name Uppercas
4. Introduction to Data Exploration and Visualization Introductory remarks The handout series are collections of 1 illustrative examples shown and discussed during the formal presentation meant to be annotated i e not always self explanatory 2 information on how to use the EDA software 3 additional examples and implicitely or explicitely suggested directions for your exploration 4 background information Example collection Ex 1 26 cantons Stem leaf ALPS 1 Initiative of the Alps rail transit Legend 216 stands for 25 51 818 for 87 54 6 668 458 122355567799 000124 I O1 amp N 8 Ex 2 183 countries Stem leaf Pop93 3 Population 1993 Legend 0 0 stands for 2000 00 1119 for 1188628990 00 0000000000000000000000000000000000000 138 2236 06 H L J O O1 amp NH mre Ex 3 Stemleaf Pop93 3 Population 1993 Legend 0 0 stands for 2000 00 34 2 for 35212000 00 0 2 NNPPPEHE NOODBRNOWDAS 0000000001111111111111222222222233344 27 144456789123335557 0113356601122369 26955778 5566779017899 1457934 007 0028 5688 294 16112 736 1334 79 4 2 27 1 E Horber 13 12 98 intro mss This example shows the default display for the same data shown in the previous example Observations much bigger or smaller relatively speaking than the others appear on a separate high labelled hi or low labelled lo stem As a
5. at this is the count of observations on the stem and not some strangely labelled observation or a stem containing digit leaves 3 In the EDA Software these names are called CASIDs EDA 1 2 amp denotes fractional leaves Stemleaf plots can be adapted for other purposes for instance comparison of the distributions of two variables on the same display showing them back to back 30 countries Stemleaf LifeEM 6 Life Expectancy with LifeEF 7 Life Expectancy Legend 68 0 stands for 68 00 82 ifeEM lo 000000 68 00 70 000000000 72 00000000000001 74 76 78 80 82 or to study differences between groups Stemleaf GNPAgr 20 Groups defined by Continents Legend Asia 0111112234 55778 01 6899 123 5567 24 9 12 LT OU EE DONNE hi 4 0 0 stands for 0 00 A frica 34 55567 1224444 566 111112 577 011344 567 3444 5555677 01123 5 GNEQ men women 0 for 83 00 LifeEF TURQ 0 0000 00000000 000000000000 0000 SGNP for Agriculture 515 for 55 00 Europe 1123333334444 55666788 134 667 03 3 The next example is a histogram showing case ids as leaves 30 countries Histogram Urb 5 Urbani midpoint 32 50 PORT 37 50 ALBA 42 50 47 50 52 50 ROUM 57 50 A IRL 62 50 TURQ CH 67 50 BULG 72 50 N I 77 50 CHE LUX 82 50 LIE S 87 50 DK MAL 92 50 ISLA D 97 50 B MON zatio
6. ata sets read by GET are EDA specific system files i e the only software package that can read and produce them is EDA Of course EDA has a number of commands to bring in data from the outside world namely the READ command and its many options But start to learn how to work with EDA using the various data sets which are readily available 6 Note that this is an EDA command and NOT the DOS DIR command EDA 1 7 STEMLEAF v lt opt gt STEMLEAF v BYGVAR gvar NGROUPS ng lt opt gt STEMLAEF v SPLIT log expression PARALLEL lt opt gt STEMLEAF vl v2 lt opt gt lt opt gt SCALE value WIDTH chars NOLINE NOHILOSTEM ASCENDING DESCENDING There are four different forms producing variations of the stem and leaf plot of the command each of them sharing a number of common options A number of metasymbols are used v Refers to a single variable Used to indicate an option Options within options Select one alternatives In the ASC DESCENDING example select either ASC or DESC if you use this option option lt opt gt see definition of lt opt gt elsewhere usually below Even though syntax diagrams might look complex sometimes frightening make sure to understand that the actual command you are typing will often be very simple e g STEMLEAF 1 sometimes with an option or two A first list of commands The
7. e because this is an acutal command line example gt BOXPLOT 1 gt BOXPLOT 1 2 4 gt BOXPLOT 1 10 PARALL gt BOXP 1 10 PAR GI HH The four examples produce boxplots The first example displays a boxplot for variable number 1 the second three boxplots for variables 1 2 and 4 Thee third example produces parallel boxplots for all variables from 1 to 10 PARALLEL is an option The last example is identical to the third except that it shows that you need not type all letters 1 1 2 4 and 1 10 show various forms of variable lists Variable lists are always specified immediately after the command name before any option Data in EDA Data you want to analyse has to be brought into the EDA work area i e the active data matrix data sheet The GET command reads a data set into the EDA Work Area WA i e the data matrix to be analysed Use the DIR command to see a list of available datasets This command will show the name and a short description of all datasets in the EDA library i e the data sets available with a GET command Syntax conventions The user s manual and the on line help use a number of syntactical conventions If you type gt STEMLEAF you will see the syntax of the STEMLEAF command Do not worry if you do not understand all the details of the command itself concentrate on the syntactical constructs used 4 Later we will learn that case and variable names are case sensitive 5 The d
8. ee the next screenfull The are some situations however where the information quickly scrolls off the screen and when the screen stops you are looking at the bottom of the display In this situation you might use the lt PAUSE gt or lt SCROLL LOCK gt keys on your PC to stop scrolling or you might tell EDA to stop after each screenfull of information this is done with the SET PAGE ON command turns paging on SET PAGE OFF turns it off 7 Metasymbols are symbols used to explain the syntax and are not used in actual commands 8 You are also offered the choice to stop at that point EDA 1 8 Additional information Type INFO INFO to see what other course specific on line information is available Basic information command lists general concepts etc can be obtained from the HELP command syntactical information on a specific command is produced by lt name gt where name is the name of a valid EDA command EDA 1 9
9. n A GREC HNGR POLO CHYP F T UK NL A FI AND The next series of examples shows various numerical summaries EDA 1 3 Ex 7 Ex 8 N amp C Am 1122233444 5667899 134 666999 124 03 Ex 9 Ex 10 183 countries Summary GNPCap 19 GNP per capita 1622 00 H 479 50 6491 50 O 71 00 50000 00 This is a 5 number summary showing the median 1622 as well as the hinges labelled H letter value and the minimum maximum labelled O for One depth 1 Ex 11 183 countries Summary GNPCap 19 GNP per capita 1622 00 spread mid H 479 50 6491 50 6012 00 3485 50 O 71 00 50000 00 49929 00 25035 50 Trimean 2553 75 Ex 12 183 countries Summary GNPCap 19 GNP per capita 1622 00 spread mid 479 50 6491 50 6012 00 3485 50 283 50 15137 50 14854 00 7710 50 191 00 21407 00 21216 00 10799 00 172 00 23383 50 23211 50 11777 75 117 00 25948 50 25831 50 13032475 84 00 30304 00 30220 00 15194 00 71 00 50000 00 49929 00 25035 50 Trimean 2553 75 Orwaovnta The next series shows boxplots starting with an example illustrating the various forms boxplots can take Ex 13 EDA 1 4
10. se commands perform common tasks and are useful to learn about exploratory tools All of them are straightforward to use and to understand from the output they produce You are invited to try them out GET name Gets a work area from the archive library DIR Shows the work areas in the archive library DESCRIBE vlist display variable info labels and descriptors DESCRIBE ALL display variable info for all variables in the WA STEMLEAF produces a stem and leaf plot HISTOGRAM shows a histogram HISTOGRAM vlist BAR classical histogram LIST listing variables many options coded etc SHOW conditional lists SHOW FAR shows only outliers BOXPLOT displays a box and whisker plot PARALLEL parallel boxplot SUMMARY numerical summaries 5 number summaries etc DISPLAY numerical summaries MEDIAN MEAN etc QSUMMARY quick summaries DLINE density lines single line histograms CODED coded density lines PLOT plot two or more variables many forms PI plot inspect module Controlling screen output Most commands produce output in a way that you can see all information on a single screen There are however exceptions output from commands producing lists usually does not fit on a single screen Commands like the LIST or DIR command will by default automatically page the output i e after a screenfull of output the display stops and you are invited to hit the return key to s
Download Pdf Manuals
Related Search