Research On The Application Of Neural Network To Credit Risk Evaluation System For Commercial Banks

The development of computer and network technology bring us into an information age, at the same time, it also make us have to face a series of problems, among them are the huge quantity of data and the complicated relations that exist in the data. How to make complete analysis to the vast data accumulated in the process of information, and how to discover hidden and valuable knowledge from the data is a hot and urgent problem for all the application fields. Data mining technology emerges as the times require and gets wide application in many fields soon, data mining technology is of more and more important value.With the increasing research and application of data mining in all kinds of fields, many excellent and practical data mining technologies come into being. Neural network is one of the most popular data mining technologies, and it is an intelligent data analyzing system constructed on the basis of simulating the structure and function of human neural cell group which has learning characteristic. Neural network has a strong ability of self-organizing, self-learning and self-adapting, and is suitable for the requirements of complex condition and multitask controlling, in addition, neural network has the characteristic of approaching any non-linear and continuous function at any precision. Generally, neural network is fit for processing non-linear data and data with noise, and it will be widely adopted in credit risk evaluating of commercial bank.Credit risk evaluation is an traditional and ancient subject, and it has a long history of development in abroad, and the method of measuring credit risk has transformed from qualitative data analysis to quantitative data analysis, and most important, modeling methods are begin to used in credit risk evaluation recently. In our country, owing to the later research on credit risk evaluation, the technique used in credit risk evaluation is still in its primary stage, and can't meet the demands of commercial banks to deal with the strong market competition in international finance area. So it is of great importance to quicken the research on credit risk evaluation and to find out the credit risk evaluation methods and models which are suit for the condition of commercial banks in china, so as to provide reliable gist for commercial banks in its decision of loan, and to make the credit risk management of commercial banks more reasonable, more scientific and more effective, and to improve the power of competition of commercial banks in the end.Neural network has its particular advantages in solving credit risk evaluation problems, most important of which is non-linear mapping ability of neural network. By forward neural network, non-linear mapping relation lies in credit data and credit grades can be easily realized, so as to achieve the function of classify customer in credit grade by credit data finally. Based on the research of credit risk evaluation actuality and application characteristic of commercial banks, combining characteristics of all kinds of data mining and neural network technologies, a credit risk evaluation model based on BP network is proposed in this paper, and then the evaluation model was improved and validated in the article. The model proposed in this paper provides a feasible scheme for commercial bank in credit risk evaluating. Concretely, this paper deal with the research in the aspect as follows:(1)To introduce the knowledge of credit risk evaluation, data mining and neural network, and to show the whole knowledge system for the research in a clear structure.