Classification Issues and Model Diagnostics

Duration: 0.5 days

This seminar focuses on issues related to selecting a cutoff when doing classification modeling. Several methods for choosing a cutoff for decision making are discussed.

Who should attend: Predictive modelers and data miners who must set or advise decision makers on setting cutoffs (used when deciding what actions to take using prediction-based models)

Prerequisites

Before attending this seminar, you should have

• general knowledge of linear statistical models
• some experience doing predictive modeling.

Course Contents

Classification Tables
ROC Curves and Measures
Direct Marketing Derived Measures: Gains Table, K-S Statistic
Cutoff, High Noon of Decision Making

• ROC and cutoff probabilities
• cutoff decisions based on event precision
• practical approach: profit

This page was created using SAS software.