Using Rough Sets Theory to Predict The Probabilities Of Credit Clients Going Bankrupt .

• Main Authors: Chuang Yi Chung, 莊宜娟
• Other Authors: 王派洲
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/33641395728935631713

碩士 === 南台科技大學 === 國際企業系 === 94 === The purpose of this study is to predict the probabilities of credit clients going bankrupt by using clients’ financial and other relevant information. We expect to decrease the credit risk and increase profits and good performance. Most of the researches adopted statistic prediction model to predict credit risk, including factor analysis, regression, and discriminate analysis. This study tried to apply the rough sets theory as the research method. It could systematically narrow down the information we need and come out with regulations. Moreover, it can develop prediction model without matching the assumption of general statistical analysis, and we use this prediction model to compare with discriminate analysis and logit regression respectively. The empirical results showed that that the rough set theory reached better result in predicting credit risk than discriminate analysis and logit regression. Applying the rough set theory, 94％can be detected one year before clients break the contracts; 86% for two years. Therefore, we can identify that the Rough Set Theory will not have the same problems as traditional statistic model when limited information was offered. In conclusion, this method is appropriate to applied in predicting the credit risk.