Macro Designing and Comparative Evaluation of Various Predictive Modeling Techniques of Credit Card Data

Abstract

Credit Scoring studies are very important for any financial house. Both traditional statistical and modern data mining/machine learning tools have been evaluated in the credit scoring problem. Predictive modeling defaulter risk is one of the important problems in credit risk management. There are quite a few aggregate models and data driven models available in literature But very few of the studies facilitate the comparison of majority of the commonly employed tools in single comprehensive study. Additionally no study assesses the performance on more then two data sets and reports the results at the same time. So a macro or a simulator is designed which would work on multiple data sets and make the process of credit scoring transparent to the novice user. In initial stage, tools were compared using Dtreg predictive modeling software. Subsequently a SAS macro is developed to evaluate the effectiveness of tools available in SAS enterprise miner. The results revealed that support vector machine and genetic programming are superior tools for the purpose of classifying the loan applicant as their misclassification rates were least as compared to others. Also cross validation is essential, though some of the tools may not support it directly.