| With the accelerating world economic integration process, there are numerousserious issues emerged in the risk exposure control of commercial bank. The level ofpersonal credit which is the essential part of bank assets business is related to thegeneral requirements of economy capital. Therefore, effective recognition andmeasurement for personal risk will become the key factor of reasonable risk controlfor the commercial bank.First, we summarize related research on single model and two-stage modeloverseas and domestic research status.the traditional classical statistical model isunstable and original artificial intelligence model is imprecise,These issuesseriously confined the development of personal credit evaluation. on the one side,An index system was built which perform well for China’s commercial banksevaluation system, on the other side, ART single model was used into personal creditevaluation field and then compares with BP neural networks, logistic regressionmodel and lineal regression model in order to analyze the advantages and drawbacksof ART model.By reduct the indexes, combine the advantage of Fuzzy set’s indexselected function and the function of deal with discrete and continuous variablesabout ART in order to achieve good classification effect. The results of usingtwo-stage model show that the optimization model improves the results to a largeextent for fuzzy classification problems and sample identification problems b y fuzzyset-ART model which has lower rate of the first and second error than ART. It meansthat the combination of two-stage model has good applicability for personal creditassessment. |
