Building Scorecards using SAS® Enterprise Miner^TM

Description

This course looks at using SAS Enterprise Miner to build credit scoring models for the consumer credit industry. It will cover the credit scoring process from the initial stage of reading the data into SAS Enterprise Miner, through the scorecard processes, performing reject inference, and validating and implementing the scorecard.

Objectives

After attending this course, you will be able to:

• identify appropriate data resources for credit scorecards
• prepare and explore data for credit scoring analysis
• build a statistical model that will use weights of evidence coding as its inputs
• develop a credit scorecard using scaling techniques
• understand the methods that can be used for reject inference
• validate the scorecard
• monitor scorecard performance

Prerequisite Skills

Before attending this course you should have:

• an understanding of statistical concepts. This knowledge can be gained by attending an introductory statistics course, for example, Applying Statistical Concepts using SAS.
• the ability to use SAS Enterprise Miner. This knowledge can be gained by attending the Applying Data Mining Techniques using Enterprise Miner 5 course.

SAS Modules Used

This course uses Version 5.2 of SAS Enterprise Miner and SAS Credit Scoring.

Course Topics

Introduction to Credit Scorecard Design

• Methodology for developing and assessing scorecards
• Preparing and exploring data for scorecard building
• The process of classing and the calculation of weights of evidence
• The use of the Interactive Grouping Node in SAS Enterprise Miner
• Building a Preliminary Credit Scorecard
• Building a logistic regression model
• Scorecard scaling using the Scorecard Node
• Assessment of preliminary scorecard
• Reject Inference
• The need for reject inference
• Three types of reject inference
• Using the Reject Inference Node to perform hard cut off augmentation, fuzzy augmentation, and parcelling
• Use of swap sets
• Comparison of methods

Scorecard Shaping

• Modifying weights of evidence groupings interactively
• Selecting variables
• Using the Regression Node to build a broadbased scorecard

Scorecard Validation

• Use of holdout sample
• Methods to assess discrimination
• Methods to assess calibration

Scorecard Management Reports

• The assessment of the scorecard
• The use of the Scorecard Node to analyse credit scores: Gains table, trade-off charts and characteristic reports
• Methods for setting the cutoff

Scorecard Tracking and Monitoring

• System stability reports
• Characteristics analysis