Research On Screening Model Of SME Credit Access To Commercial Bank

In recent years, as the development of SMEs faster and faster and the position in the national economy becoming more and more important, financing difficulties of SMEs have attracted wide attention. According to the analysis of reality, the banks play a very important role in resolving this issue. Due to the disadvantages of SMEs, it means that there is great risk if the banks develop SME credit. However, with changes in economic environment and national policy-oriented, banks are fully aware of the advantages and necessity of the development of SME credit. Then controlling the credit risk of SMEs has become a major problem in the development of SME credit. In the article, the article has done some exploratory research in the control of SME credit risk by the way of constructing the screening model of SME credit access to commercial bank. We hope it can play a guiding role in the theory.Firstly, on the basis of a large number of domestic and foreign literature, the article has summarized them in four aspects, which has laid theoretical basis to the below of the article. Secondly, the in-depth study in the current status of SME credit and SME credit mechanism of Chinese commercial banks, the article has concluded the development of bank SME credit is a general trend. However, because of the high risk of SME credit, banks must take the right model for the development of SME credit. Based on the current credit mechanism problems, the article has given certain improvements. Thirdly, based on studying the principle of the neural network method and Bayesian network method, the article has analyzed the feasibility in building models of the two methods, and then compared the two methods from several aspects. Finally, according to certain principles the article has selected index system and then builds the screening model of SME credit access to commercial bank based on neural network and Bayesian network. Then the article trains and tests the models using actual data of banks, then obtains the satisfying results.