The Research On Retail Customers Credit Evaluation Of Commercial Banks Based On The BP Neural Network

Credit is the basis of market economy. National cr edit, corporation credit and personal credit become common to build a complete credit system. Improving the credit system has a very important significance to promote the healthy operation of the economy. Personal credit system is the cornerstone of the so cial credit system. Improving the personal credit system is to establish a social market economic order and prevent financial risks effectively guaranteed. Good personal credit system can promote the rapid development of consumer credit, also expand domestic demand and stimulate economic growth. With the further development of China's consumer credit, personal credit assessment draws more and more theoretical and academic attention. Objective comprehensive evaluation of personal credit provide s scientific, timely and effective decision basis for commercial banks to make consumer credit and develop personal financial business. It can effectively reduce the credit risk, and improve the efficiency of credit decisions.Evaluation of personal credit is an important practice in the field of the current financial problems to be solved, also a serious and urgent topic issue faced with academic community. Since China started to build credit system late, the society lacked in relevant information of individual credit assessments, professional researchers and assessment agencies, which causes there is no relative unified reasonable personal credit scoring system. Therefore, to study commercial bank s retail customer credit evaluation has important practical significance.This article attempts to use BP artificial neural network model in Chinese Commercial Banking Retail Customer Credit Appraisal and practice, focusing on the BP neural network model in retail customer credit risk evaluation system. Firstly, this paper describes the personal credit and related theories and methods of personal credit evaluation, and highlights the BP neural network basic principles and operational processes, and its applicability in commercial bank credit rating. Then it uses the BP neural network self-learning ability, nonlinear processing ability and fault tolerance, to build the retail customer credit evaluation model after determining the personal credit evaluation index parameters. And the model obtained good prediction effect in the implementation of simulation by MATLAB software. But considering the BP neural network may fall into local optimum value, during training right networks and thresholds. Therefore this article proposes genetic algorithm with initial weights and thresholds of BP neural network optimization. The results showed the mean square error of optimization model is smaller using MATLAB software simulation optimization model for the empirical analysis, and the prediction precision is higher.