| Summary: | 碩士 === 國立交通大學 === 工業工程與管理系 === 90 === Credit rating is used for bank to investigate the repayment ability of borrower. Because of economics depression and higher overdue loan rate, more and more bank start to improve their present credit rating model to be a batter one. Character, Capacity, Capital, Collateral, and Condition of business are the factors that impact the borrower’s credit. And how to use these factors to built a useful credit rating model is just the importance that bank want to know.
Based on the above, we should use logistic regression to build a clear procedure for bank to construct a batter credit rating model. The procedure has three stages：（1）variables selection and data collection, （2）building risk assessment model for bank loan （binary discrimination）,（3）constructing multi-level credit rating model. Among that, risk assessment model is used to classify borrowers into regular group and default group and credit rating model is used for bank as a multi-level discrimination to investigate borrowers’ credit more flexibly.
Finally a finance company is chosen as a example in the study. We use above procedure to construct this company’s credit rating model. Our finding shows that not only risk assessment model for binary classification has high correct rate but using multi-level credit rating model we could find out the default borrowers which have shortest payment life（under 3 months）. According to above, it shows the model constructing by this procedure is useful for the finance company. Another bank and finance company also can follow this procedure to build their own credit rating model.
|
|---|
