Preface

IBM SPSS Modeler is the most comprehensive workbench-style data mining software package. Many of its individual modeling algorithms are available elsewhere, but Modeler has features that are helpful throughout all the phases of the independent, influential Cross Industry Standard Practice for Data Mining (CRISP-DM). Considered the de facto standard, it provides a skeleton structure for the IBM SPSS Modeler Cookbook and the recipes in this book will help you maximize your use of Modeler's tools for ETL, data preparation, modeling, and deployment.

In this book, we will emphasize the CRISP-DM phases that you are likely to address working with Modeler. Other phases, while mentioned, will not be the focus. For instance, the critical business understanding phase is primarily not a software phase. A rich discussion of this phase is included in the Appendix, Business Understanding. Also, the deployment and monitoring phases get a fraction of the attention that data preparation and modeling get because the former are phases whereas Modeler is the critical component.

These recipes will address:

• Nonobvious applications of the basics
• Tricky operations, work-arounds, and nondocumented shortcuts
• Best practices for key operations as done by power users
• Operations that are not available through standard approaches, using scripting, in a chapter dedicated to Modeler scripting recipes

While it assumes it will provide you with the level of knowledge one would gain from an introductory course or by working with user's guides, it will take you well beyond that. It will be valuable from the first time you are the lead on a Modeler project but will offer much wisdom even if you are a veteran user. Each of the authors has a decade (or two, or more) of experience; collectively they cover the gamut of data mining practice in general, and specifically knowledge of Modeler.