| The management of credit risk of commercial banks has been a focus of thecademiccommunity and application of commercial banks for long. The safety of commercial banksis related to the whole economics of a country because of its special position and importantrole. Because of the complexity and Nonlinear characteristics about credit risk, many riskassessment systems can not adapt to it so that they can not evaluate it accurately. NeuralNetwork Model has the ability of approximating the nonlinear function, recognizing thepattern and generalizing and classifying, so it can simulate the assessment of credit riskwell, which is a new development of the traditional method of credit risk.In this article, we analyzed the demand to establish an advanced credit riskmanagement method home an abroad first and discussed the feasibility of using neuralnetworks to simulate the credit risk. Then, the author summarized the previousachievement about the credit risk of commercial banks domestic and abroad. Moreover, theauthor summarized the advantages and disadvantages of commercial bank credit riskmodel. On the base, the author got the direction of further study. Secondly, the author gavea comparative analysis of the commercial bank credit risk management models currently,from which the author extracted the advantages that can be used for credit risk models ofneural networks. Finally, by using the conclusions of previous studies, the author put theAdaboost algorithm into the BP neural network, constructed of the optimized neuralnetwork credit risk models based on BP-Adaboost strong classifier and realized thesimulation of the assessment about financial data o f one hundred and fifty listedcompanies. From the result, the author achieved the strengths and weaknesses about theoptimized neural network credit risk models based on BP-Adaboost strong classifier. Thenthe author discussed the practicality of this model for commercial bank credit riskaccording to the empirical results. |
PDF Full Text Request