The Application of Rough Set And Factor Analysis in Default Risk Evaluation Model of Credit Card

• Main Authors: Shu Fan, 劉書汎
• Other Authors: Tsung-Nan Chou
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/08370495314275079836

碩士 === 朝陽科技大學 === 財務金融系碩士班 === 97 === The aim of this study is to develop sophisticated default risk evaluation models for credit card. The empirical investigation is divided into two sections. The first section is to construct two models including Back Propagation Network (BPN) and Logistic Regression (LR) to predict the default cases. In the BPN model, we group variables into four categories with various feature extraction methods prior to perform the comparison of the predictive accuracy. Empirical results show that BPN model based on the feature extraction of rough set achieves higher adaptability and prediction accuracy. In the second section, the rough set approach is compared with the rule based classifier, rule based classifier with discrete data, decomposition tree, k Nearest Neighbor (k-NN) classifier, and Local Transfer Function Classifier. Empirical results show that the rule based classifier with Learnable Evolution Model (LEM Algorithm) is superior to other approaches.