Study On Customer Risk Evaluation Model For Post Savings Bank Credit Based On Data Mining

With the rapid development of financial market, its security problems have become increasingly apparent. Establish risk control technology of our country’s banks, to support commercial banks to expand the new profit point of growth and steady development of the business of credit card. Although there are many for the research of credit risk, but because of the difference of specific customer groups and the postal savings bank credit card management norms of sex, the existing model and method is not suitable to risk assessment and management in postal savings bank credit card. This paper aims to make the necessary improvements to existing credit scoring model, put forward a risk evaluation model with the postal savings bank credit card customer characteristics.The main contents are as follows:First, establish the entry qualifications and compliance risk rating of two personal credit card index system based on analysis of the characteristics of the postal savings bank customers, and apply the weighted sum of squares method to determine the weighting factor index entries, by testing shows that both the index system the audit results are in line with the actual situation, the validity and applicability; the second is a combination of the index system, the use of BP neural network algorithm to optimize the existing postal savings bank credit risk assessment model based on actual customer data for the Postal Savings Bank sample-based classification and regression tree analysis, Bayes discriminant analysis, neural network model to compare credit assessment test, after test results show that the improved neural network predictive credit scoring model has a high accuracy rate, there are some practical applications value.