Develop Predictive Models Using Your Business Expertise
Contents
1. ponent with a new Predictive Applications Wizard In addition Clementine integrates with Microsoft Analysis Services for building decision trees IBM DB2 Data Warehouse Enterprise Edition for deploying models to Intelligent Miner Scoring and viewing Clementine models with Intelligent Miner Visualization m On Sun Solaris systems Clementine supports Kerberos security a network authentication protocol E Server 8 Clementine Server E Client Clementine integrates with SPSS for Windows and Answertree and supports Cleo through a built in Wizard a It also integrates with other databases and market automation and software tools System requirements Clementine Client Operating system Windows XP Home Edition Windows XP Professional Windows 2000 Professional Windows Server 2003 Windows 2000 Server or Windows NT 4 0 with Service Pack 6 Memory 512MB RAM recommended a Minimum free disk space 320MB A CD ROM drive is required for installation Web browser Internet Explorer version 5 x or later or Netscape 6 or later for running the online Help system Clementine Server Clementine Solution Publisher Runtime and Clementine Batch Operating system Windows Server 2003 Windows 2000 Server Windows 2000 Professional or Windows NT 4 0 with Service Pack 6 or later Solaris 7 8 or 9 HP UX 11i IBM AIX 5 2 or OS 400 on the IBM eServer iSeries V5R2 with OS 400 Portable App2. Statistical output Clustering and segmentation Kohonen network K means and TwoStep View cluster characteristics with a graphical viewer Association detection GRI Apriori sequence CARMA Filter sort and subset association models with the new association model viewer Use Apriori sequence and CARMA generated models for scoring Data reduction Factor analysis and principal components analysis View model equation and advanced Statistical output a Combine models for greater accuracy through meta modeling Multiple models can be combined or one model can be used directly for building a second model a Import PMML generated models created in other tools such as AnswerTree and SPSS for Windows Clementine External Module Interface CEMI for custom algorithms Add data access modeling and graphics Create node palettes and generated model palettes including PMML generated models created in other tools using custom CEMI icons a An algorithm user manual is included which explains the theories and methods behind the algorithms included in Clementine Evaluation a Easily evaluate models using lift gains profit and response graphs Use a one step process that shortens project time when evaluating multiple models Define hit conditions and scoring expressions to interpret model performance a Analyze overall model accuracy with coincidence matrices and other automati
3. it can process it passes only the relevant aggregated tables to Clementine Server If you are using a CEMI such as Text Mining for Clementine or Capri Clementine Server passes the relevant tasks to that particular external process Features are subject to change based on the final product release _ Symbol indicates a new feature Separately priced modules Cleo Web server Cleo requires at least one server computer that meets the following minimum requirements Using additional processors faster processors and more RAM will improve performance Operating system Windows 2000 Server or Windows Server 2003 Windows NT 4 0 Server with Service Pack 5 or higher cannot be installed on Windows NT Terminal Server or Solaris Version 7 or later Hardware Pentium compatible processor 500Mtz or faster if running on Windows Sun UltraSPARC II or better for Solaris Memory 512MB RAM Minimum free disk space 7OOMB A graphics adapter with 800 x 600 resolution SVGA or higher capable of displaying at least 256 colors Anetwork adapter running the TCP IP protocol Repository the system requires a database to serve as a repository for published content framework settings and other information The following databases are supported Microsoft SQL Server 2000 Oracle 8i version 8 1 7 Data warehouse the system can be configured to access data from a data warehouse or database The s
4. Clementine 8 5 Specifications gt Develop Predictive Models Using Your Business Expertise Clementine is an integrated data mining workbench popular worldwide with data miners and business analysts alike It supports the entire data mining process enabling users to quickly access and prepare data for modeling rapidly build and compare models and then deploy models into real time operational environments And it seamlessly integrates data mining results with other business systems and processes giving decision makers a clearer view of current conditions and insight into future events Clementine s highly visual user interface makes train of thought analysis possible Clementine produces streams a visual map of your data mining steps to automatically document the analytical process By interacting with a stream data miners can add business knowledge the key to successful data mining to the data mining process The Clementine interface makes it easy to explore ideas or intuitive hunches And because it allows you to focus on knowledge discovery rather than on performing technical tasks such as writing code you achieve results quickly Select from an unparalleled breadth of techniques Clementine offers a full range of machine learning and statistical techniques for clustering classification association and prediction SPSS Inc has more than 35 years experience in predictive analytics and its algorit
5. a Features CRISP DM Help your company s analysts and data miners focus on business problem solving rather than on the data mining technology Clementine supports the CRoss Industry Standard Process for Data Mining CRISP DM the de facto industry standard at every step of the process Clementine projects can be efficiently organized using the CRISP DM project manager Business lt Understanding lt Data Understanding The CRISP DM process as shown above enables data miners to efficiently implement data mining projects that yield measurable business results Business understanding Clementine s visual interface makes it easy for your data miners to apply business knowledge to data mining projects In addition optional business specific Clementine Application Templates CATs are available accelerate the attainment of data mining results CATs ship with sample data that can be installed as flat files or as tables in a relational database schema You may choose from the a CRM CAT Telco CAT a Fraud CAT Microarray CAT Web mining CAT requires the purchase of Web Mining for Clementine Data understanding You can obtain a comprehensive first look at your data through Clementine s data audit node Create summary Statistics histograms and distribution graphics for each data field and display them in an easy to read matrix Clementine also enables you to visually inter
6. act with output by generating subsets of data from graphs and tables You can also Select a region of a graphic and view the selected information in a table or use this information downstream Create histograms distributions line plots and point plots Use Web association detection Display 3 D panel and animated graphs View data quickly through graphs summary statistics or an assessment of data quality 29 Data preparation Access data Structured tabular data Manage data security through the selected database and or operating system Access ODBC compliant data sources with the SPSS Data Access Pack which ships with Clementine Drivers in this middleware pack include support for Oracle Microsoft SQL Server IBM DB2 Informix and Sybase databases Import delimited and fixed width text as well as any SPSS and SAS 6 7 8 and 9 file Unstructured textual data Automatically extract concepts from documents and from text notes in databases using Text Mining for Clementine Web site data Automatically extract Web site events from Web logs using Web Mining for Clementine Data output Work with delimited and fixed width text ODBC Microsoft Excel SPSS and SAS 6 7 8 and 9 files Choose from various data cleaning options Remove or replace invalid data Automatically fill in missing values Manipulate data Work w
7. c evaluation tools Deployment Clementine offers a broad array of deployment capabilities to meet your organization s needs Models built in Clementine can be directly deployed into other SPSS predictive applications and into other vendors technologies Clementine Solution Publisher optional Automate the export of all operations including data access data manipulations model scoring including combinations of models and post processing Workin a runtime environment for executing image files on target platforms Easily update solutions through small image files a PredictiveCallCenter optional Automatically export full Clementine streams to the Integration Builder component of PredictiveCallCenter Combine exported Clementine streams with PredictiveCallCenter models business rules and exclusions to optimize customer interactions Cleo optional A Web based solution for rapid model deployment Enable multiple users to simultaneously access and immediately score single records multiple records or an entire database through a customizable browser based interface a Clementine Batch Automate production tasks while working outside the user interface Automate Clementine processes from other applications or scheduling systems Generate encoded passwords Call Clementine processes via the command line a Scripting Use command line scripts or scripts assoc
8. hms are calibrated and verified to support the creation of powerful data mining models With Clementine you can be assured of having the analytical techniques you need to attain the most reliable results Clementine 8 5 incorporates CARMA a high speed association detection algorithm and offers a new association viewer Integrate with existing information systems With Clementine there s good news for your IT staff Clementine is an open standards based solution that easily integrates with your organization s existing operational systems In addition Clementine leverages your database investment by performing much of the data preprocessing tasks typically 80 90 percent of your data mining effort within the database Because preprocessing occurs on the database tier rather than on the server or client tiers you save time maximize resources and minimize network traffic And with Clementine 8 5 organizations using Microsoft SQL Server 2000 Analysis Services can build Microsoft decision trees directly from the Clementine interface Put results to work in your organization SPSS offers several deployment options to meet your needs for in database or real time scoring These include saving models in industry standard Predictive Model Markup Language PMML and embedding them as stand alone files in other systems and applications You can also export not only models but also all data mining steps including data access
9. iated with Clementine streams to automate repetitive tasks in the user interface Scripts generally perform the same types of actions that otherwise are carried out using a mouse or keyboard Execute selected lines from a stream SuperNode or stand alone script using an icon on the toolbar Update stream parameters within a standalone script Export generated models as PMML Perform in database scoring which eliminates the need for and costs associated with transferring data or performing calculations on the client Generate SQL from linear regression and factor generated models Generate SQL from decision trees and rule sets a Use the bulk loading capabilities of your database Increase performance during data export by using your database s bulk loader Fine tune various options including row wise or column wise binding for loading via ODBC and batch size setting for batch commits to the database Bo ae Scalability and integration Scalability a Use in database mining to leverage parallel database implementations a Minimize network traffic via intelligent field projection Clementine pulls data only as needed from your data warehouse and passes only relevant results to the client Integration Clementine supports the export of streams to SPSS PredictiveCallCenter application through PredictiveCallCenter s Integration Builder component m Export streams to the Integration Builder com
10. ith complete record and field operations including Field filtering naming derivation and value replacement Record selection sampling merging and concatenation sorting aggregation and balancing deriving new fields based on conditional criteria and calculating new fields Specialized manipulations for showing the history of values and converting set variables into flag variables Automatically bin data based on n tiles mean and standard deviation rank or a fixed width Re group or collapse categories and transform one set of discrete values into another Change the order of fields a Merge records using any of the following types of joins Inner join Full outer join Partial outer join both left and right Anti join the opposite of an inner join Features are subject to change based on the final product release _ Symbol indicates a new feature Separately priced modules Modeling a Prediction and classification Neural networks multi layer perceptrons using error back propagation radial basis function and Kohonen networks Browse the importance of the predictors Decision trees and rule induction C5 0 and Classification and Regression Trees C amp RT Browse and interact with decision trees Browse collapse and expand decision rules Linear regression logistic regression and multinomial logistic regression View model equation and advanced
11. lications Solution Environment PASE 5722 SS1 Option 33 External Processing e g modeling algorithms Hardware Intel Pentium compatible processor if running on Windows Sun UltraSPARC II processor or better if running on Solaris PA RISC processor and HP 9000 server or HP Workstation if running on HP UX PowerPC 233 MHz or faster and IBM RS 6000 for AIX or IBM iSeries server for OS 400 m Hardware 64 bit processor supported on SPARC servers running Solaris 9 Memory 512MB RAM minimum a Minimum free drive space 128MB of available disk space are required for installation Additional free disk space is required to run the program for temporary files 1GB is recommended For Clementine Solution Publisher Runtime the minimum free disk space required to install the software is 64MB plus at least twice the disk space of the amount of data to be processed A network adapter running TCP IP protocol A CD ROM drive is required for installation Clementine Application Templates a Clementine 7 2 Memory 512MB RAM recommended Clementine uses a highly scalable three tiered architecture for data mining as shown in this diagram The Clementine Client tier shown at the bottom of the illustration passes stream description language SDL to Clementine Server Clementine Server then analyzes particular tasks to determine which it can push to the database After the database runs the tasks
12. modeling and post processing through Clementine Solution Publisher These and other deployment options increase the usability of Clementine models and save your organization time and money With Clementine 8 5 companies can automatically deploy models built in Clementine into IBM DB2 Data Warehouse Enterprise Edition This release also enables companies to author and deploy streams from Clementine to the PredictiveCallCenter application from SPSS New features provide additional power and integration enabling your organization to further leverage your investment in information technology If you use Microsoft SQL Server 2000 Analysis Services you can use Clementine to build Microsoft decision trees If you use Clementine on the Sun Solaris platform you will enjoy improved performance in processing large datasets because Clementine 8 5 is available as a 64 bit application on SPARC servers If you use Clementine with IBM DB2 databases you can browse association models and automatically deploy models built in Clementine into DB2 Data Warehouse Enterprise Edition If you use PredictiveCallCenter from SPSS you can author and automatically deploy Clementine streams to PredictiveCallCenter s Interaction Builder component Other new features include a new association viewer and CARMA a high speed association detection algorithm that delivers faster processing and gleans more detailed information from your dat
13. uirements a Operating system Windows XP Home Edition Windows XP Professional Windows 2000 Professional with Service Pack 2 or later Windows Server 2003 or Windows 2000 Server a Memory 512MB RAM Minimum free disk space twice the amount of raw Web data being processed Capri for Clementine Plug in algorithm for detecting sequences Operating system Windows 98 Windows 2000 Professional Windows Server 2003 Windows 2000 Server Windows NT 4 0 with Service Pack 3 or later or Solaris 6 Hardware Pentium compatible processor if running on Windows or SPARC for Solaris a Memory 128MB RAM Minimum free drive space 3MB A CD ROM drive is required for installation
14. ystem has only been tested with SQL Server 2000 and Oracle 8i databases Web client content is delivered to clients as Standard HTML pages Supported browsers include Internet Explorer version 5 5 with Service Pack 2 or version 6 0 for Windows Internet Explorer version 5 2 for Macintosh Netscape 6 2 Text Mining for Clementine Client version requirements Clementine 7 1 or later Operating system Windows XP Professional Windows 2000 Professional Windows NT 4 0 with Service Pack 6 or later or Windows 2000 Server Memory 512MB RAM Minimum free disk space 85MB plus space for databases Web Browser Internet Explorer 5 0 or later or Netscape 6 0 or later is required to use the Viewer node Server version requirements Operating system Windows 2000 Server Windows NT Server 4 0 with Service Pack 6 or later Solaris 8 or 9 Hardware Pentium III processor 1GHz or faster if running on Windows Sun UltraSPARC II or better if running on Solaris Memory 256MB RAM Minimum free disk space 85MB plus space for databases Web Mining for Clementine Client version requirements a Clementine 7 2 or later a Operating system Windows XP Home Edition Windows XP Professional Windows 2000 Professional with Service Pack 2 or later Windows Server 2003 or Windows 2000 Server a Memory 512MB RAM Minimum free disk space twice the amount of raw Web data being processed Server version req
Download Pdf Manuals
Related Search