C5.0 Classification Algorithm and Application on Individual Credit Evaluation of Banks

Abstract

This article focuses on individual credit evaluation of commercial bank. The records of individual credit include both numerical and nonnumeric data. Decision tree is a good solution for this kind of issue. This year, the algorithm C4.5 of decision tree become popular, but C5.0 algorithm is still undergoing. In this article, we do some deep research on C5.0 algorithm by embedding “boosting” technology in cost matrix and cost-sensitive tree to establish a new model for individual credit evaluation of Commercial Bank. We apply our new model on evaluating the individual credit records of a German bank, and compared results of the adjusted decision tree model and the original one. The comparison shows that the adjusted decision tree model is more precise.