Study Of Intelligent Credit Risk Early Warning Method

For a long time, the management of credit risk always is a focal problem attended by the money market. Skills about it, especially based on statistical analysis, have been matured in the west countries, and many tools and software has been made for the application of commerce. However, because of its restrict and disadvantages the statistical analysis couldn't have good performance on the application in China. Therefore, other approaches have to be considered.In this paper, first basic concepts in the field of credit risk are introduced, then some chief credit evaluation models based on statistical analysis are discussed, and disadvantages of them are pointed out. This paper advanced a new method, which is the combination of Back propagation Learning and the Self-organizing Map, to evaluate credit risk. Then the structure and the arithmetic of credit evaluation model is discussed.Second, this paper uses the financial data of companied on the market publicized in Shanghai 2005 as sample set. After these samples have been predisposed, the model is trained and tested, and the performance of it has been analyzed.Last, this paper designs and develops prototype system for credit Evaluation and risk early warning. In the system, rule for credit can be defined and managed neatly, which improved the currency of the system. Furthermore, interface for data exchange is emphasized.The research of above can be applied in the design and development of bank risk evaluation early warning system.