Overview

This course covers predictive modeling using SAS/STAT software with emphasis on the LOGISTIC procedure.

This course also discusses selecting variables and interactions, recoding categorical variables based on the smooth weight of evidence, assessing models, treating missing values, and using efficiency techniques for massive data sets.

Who should attend:

Modelers, analysts, and statisticians who need to build predictive models, particularly models from the banking, financial services, direct marketing, insurance, and telecommunications industries

Prerequisites

Before attending this course, you should:

  • Have experience executing SAS programs and creating SAS data sets, which you can gain from the SAS Programming 2: Data Manipulation Techniques course.
  • Have experience building statistical models using SAS software.
  • Have completed a statistics course that covers linear regression and logistic regression, such as the Statistics 1: Introduction to ANOVA, Regression, and Logistic Regression course..


This course addresses SAS/SHARE software.

Delegates will learn how to

Learn how to:

  • Use logistic regression to model an individual's behavior as a function of known inputs.
  • Create effect plots and odds ratio plots using ODS Statistical Graphics.
  • Handle missing data values.
  • Tackle multicollinearity in your predictors.
  • Assess model performance and compare models.

Outline

Predictive Modeling

  • Business applications.
  • Analytical challenges.


Fitting the Model

  • Parameter estimation.
  • Adjustments for oversampling.


Preparing the Input Variables

  • Missing values.
  • Categorical inputs.
  • Variable clustering.
  • Variable screening.
  • Subset selection.


Classifier Performance

  • ROC curves and lift charts.
  • Optimal cutoffs.
  • K-S statistic.
  • c statistic.
  • Profit.
  • Evaluating a series of models.