About this Course

This course demonstrates how to develop models to predict categorical and continuous outcomes, using such techniques as neural networks, decision trees, logistic regression, support vector machines, and Bayesian network models. Use of the binary classifier and numeric predictor nodes to automate model selection is included. Feature selection and detection of outliers are discussed. Expert options for each modeling node are reviewed in detail and advice is provided on when and how to use each model. You will also learn how to combine two or more models to improve prediction. Independent Study Only: Syllabus is provided for each week's study and materials are completed privately by each participant. 1 time per week students will meet on-line to review course exercises with a Live Instructor.

Audience Profile

This advanced course follows either 'Introduction to IBM SPSS Modeler and Data Mining' or Advanced Data Preparation with IBM SPSS Modeler is essential for anyone who wishes to become familiar with the full range of modeling techniques available in IBM SPSS Modeler to create predictive models.

Prerequisites

You should have:

• General computer literacy
• Experience using IBM SPSS Modeler (formerly Clementine) , including familiarity with the IBM SPSS Modeler environment, creating streams, reading in data files, assessing data quality and handling missing data (including the type and data audit nodes), basic data manipulation (including the derive and select nodes), and creation of models.
• Prior completion of Introduction to IBM SPSS Modeler and Data Mining is required and completion of Advanced Data Preparation with IBM SPSS Modeler is strongly encouraged.
• An introductory course in statistics, or equivalent experience, would be helpful for the statistics-based modeling techniques.

Course Outline

1. Preparing data for modeling
2. Searching for data anomalies
3. Selecting predictors
4. Data reduction with principal components
5. Neural networks
6. Support vector machines
7. Cox regression
8. Time series analysis
9. Decision trees
10. Linear regression
11. Logistic regression
12. Discriminant analysis
13. Bayesian networks
14. Numeric Predictor node
15. Binary Classifier Node
16. Combining models to improve performance
17. Getting the most from models
18. Appendix A: Decision List