Problem Formulation

Overview of the Predictive Modeling Case

A financial services company offers a home equity line of credit to its clients. The company has extended several thousand lines of credit in the past, and many of these accepted applicants (approximately 20%) have defaulted on their loans. By using geographic, demographic, and financial variables, the company wants to build a model to predict whether an applicant will default.

Input Data Source

After analyzing the data, the company selected a subset of 12 predictor (or input) variables to model whether each applicant defaulted. The response (or target) variable BAD indicates whether an applicant defaulted on the home equity line of credit. These variables, along with their model role, measurement level, and description are shown in the following table.

Note: This book uses uppercase for variable names. SAS accepts mixed case and lowercase variable names as well.

Name	Model Role	Measurement Level	Description
BAD	Target	Binary	A value of 1 indicates that the client defaulted on the loan or is seriously delinquent. A value of 0 indicates that the client paid off the loan.
CLAGE	Input	Interval	Age of the oldest credit line, measured in months
CLNO	Input	Interval	Number of credit lines
DEBTINC	Input	Interval	Debt-to-income ratio
DELINQ	Input	Interval	Number of delinquent credit lines
DEROG	Input	Interval	Number of major derogatory reports
JOB	Input	Nominal	Occupational categories
LOAN	Input	Interval	Amount requested for the loan
MORTDUE	Input	Interval	Amount due on the existing mortgage
NINQ	Input	Interval	Number of recent credit inquiries
REASON	Input	Binary	The value DebtCon indicates that the loan was intended for debt consolidation. The value HomeImp indicates that the loan was for home improvement.
VALUE	Input	Interval	Value of the current property
YOJ	Input	Interval	Years at the applicant’s current job

The SAMPSIO.HMEQ data set contains 5,960 observations for building and comparing competing models. The data set is split into training, validation, and test data sets for analysis.