The Research On Personal Credit Evaluation System For College Students Based On PSO-BP Model

In recent years, credit issue has already become an important topic which is focused by the whole society. Personal credit is the core and foundation of the whole credit system. The research and construction on personal credit evaluation system is particularly important. College students are the future in society, their credit viewpoint and credit status directly affects the credit level of the future society. But at present, the status about personal credit of college students is not optimistic. Therefore, it's very necessary to establish personal credit evaluation system for college students. It will help students be supervised, also will guide students to emphasize personal credit records. At the same time, it will improve college students'personal credit behavior and promote the work of Government-Subsidized Student Loans and employment, etc.The paper discussed the background and the significance of researching personal credit evaluation system for college students, and then introduced the related theory and some traditional credit evaluation methods. According to the actual situation of universities in our country, we established personal credit evaluation index system for college students through referencing other scholars'research and the method of expert evaltion. The index system included two grades. After that, in order to evaluate personal credit objectively, truly and scientifically, we processed the qualitative indexes, and set the standard of credit rating. In order to collect data, the paper designed a questionnaire survey. We undertook to students anonymously in five different universities in Shanxi province which belong to different level. At last, we got sample data, training sample data and test sample data through sampling and grouping.In this paper, we constructed a model based on neural network. But BP model has its inherent defects, such as its convergence speed is slow, the result is easy into the local minimum, and initial parameters are difficult to determine. So in this paper, we used PSO algorithm to optimize the BP neural network model. As a good optimization tool, PSO algorithm modified the weights and thresholds of BP neural network model constantly in the training process. PSO algorithm causes the optimization process no longer depend on gradient information. This new model overcomes the inherent defects of BP model.This paper used PSO-BP to construct the model of personal credit evaluating for college students. The program was written and simulated by Matlab. In order to prove the effectiveness of the model PSO- BP, sample data was trained by two models respectively, and then we made a comparison and analysis for them. The result showed that PSO accelerated the convergence speed, improved the generalization ability of BP model. According to the training effect, PSO-BP model is obviously superior to BP model. It fully demonstrated the superiority of PSO-BP model. Finally, we used PSO-BP model which was trained successfully to do empirical analysis for part of data.