The Evaluation Of Personal Housing Loan Credit System Based On BP Neural Network

With the development of our society and economy, the real estate industry is increasingly associated with people’s live. Whether business or personal, they both take the investment in real estate as an important asset investment. Therefore, personal housing loan business has become one of the largest businesses of commercial banks. And the repayment ability of businesses and individuals has become an important index of credit risk management in commercial banks.In overseas, the assessment of personal payment ability in commercial bank has achieved fruitful results. At present, it makes quantized analysis, in statistics methods, of the loan paying capability of customer. Currently, in our country, the establishment of customer evaluation system is relatively backward, and far away from a complete system. In this environment, commercial banks face greater credit risk. Consequently, in order to provide a scientific basis for commercial bank decision-making we should speed up the job of our evaluation system.Neural network is in natural of self-learning,self-regulation and nonlinear mapping. It also can quantify the model of personal credit evaluation. BP neural network as a kind of neural network is used more widely. In this paper we use the bank information of customer’s housing loan repayment records, and then select11evaluation indexes to construct evaluation index system. To design a profit BP neural network model in matlab. The input data were deal with normalization and redundancy. Besides, the research of BP neural network is thorough and its algorithms are numerous. For different algorithms, we can’t comment this one is better than the other one simply, but we can find the most suitable method for every specific question. In the design process of BP neural network model, by comparing the performance, training frequency and training time of different training algorithm, Momentum, BFGS Quasi-Newton, Conjugate Gradient with Polak-Ribiere Restarts, L-M algorithm respectively. Finally we decide to use the BP neural network which based on L-M algorithm in this paper. After the completion of the model training and learning process, we choose a new customer’s housing loans information to verify the availability of this model. By comparing the expectation and prediction of test result, it demonstrates that our method is effective. Therefore, it provides a feasible solution to personal housing credit evaluation.