Research on Credit Risk Assessment in Commercial Bank Based on Information Integration

摘要：

Credit risk is the main risk faced by commercial bank during operation, how to manage credit risk is a focus for financial specialists. Therefore effective assessment of credit risk is very important. The risks faced by banks during operation can be regarded as a kind of research of uncertain problems study. The essence of multi-sensor data integration is as follows: to have a best decision according to much information from different sensors. A kind of credit risk assessment model based on information integration is developed depending on the idea of multi-sensor information integration. This model includes three algorithms, as follows: BP neural network, SVM and DS evidence theory. This model has not only the classifying capacity of BP neural network and SVM, but also the decision ability of DS evidence theory. The essence of the model has two key steps, first we can obtain elementary decision, then we depend on DS evidence theory has a integration of processed outputs data by BP neural network and SVM. Depending on some data of certain bank, we have a simulation aiming at three models: BP model, SVM model and new developed model. The results show the developed new model can obtain better assessment compared with two models at same conditions.

关键词： Credit risk BP neural network SVM DS evidence theory

作者: GUO Yingjian WU Chong

作者单位: School of Management,Harbin Institute of Technology,P.R.China,150001

会议类型: 国际会议

会议名称: 2009 International Conference on Management Science and Engineering(2009管理科学与工程国际会议)

会议地点: 北京

会议语种:英文

页码: 515-525

在线出版日期: 2009-11-01（万方平台首次上网日期，不代表论文的发表时间）