Learning Vector Quantization Neural Networks Applied To Credit Rating

Credit rating is referred to an economic activity, which is conducted by credit rating institute. They evaluate the economic obligation implementation capability and credit ratio of a department candidly based on scientific index system, analytical method and simple accurate symbol.The development of reputation rating industry is closely related to that of a country's capital market and even to its economic development. Generally speaking, the more developed and canonical a capital market is. the more mature and flourishing the credit rating industry reaches. The target of our country's economic system reforms has already been very explicit, that is to build up a right market mechanism. Since 1997 the stock market has already been on a road of steady development after experiencing groping development phase; the function of stock market is extended and becoming more right. The significance that a stock market plays in pushing forward the reform and economic development of state owned business enterprise is well recognized by people from all state of life. Credit rating is one of the effective means the prompts the function of market mechanism. Therefore, the institute of credit rating will definitely be an effective part of market.Carrying on a credit rating to a listed company, building up associated model, and making use of this model to predict the possibility of a certain circumstances or property occurrence will make discover credit crisis signal earlier. This will help the manager take effective measures to improve management method , so as to prevent management crisis at an embryonic stage when crisis occurs, and also make investor and creditor transfer their property in time, manage accounts receivable and make a letter loan decision according to this kind of signal; Depending on that mode, also the auditor can evaluate the management condition of a business enterprise accurately and avoid litigation resulting from not canning its management failure reasonably. Meanwhile, the mode can also help reduce the cost that stock supervising department will spend on taking charge of the quality analysis a listed company. Facing those which prepare to lend "hull" or buy "hull", it is also essential to estimate financial before looking for reorganized company. Currently, the strength of listed companies varies a lot, in order to avoid those cheating means, make investors understand dynamic stated of market and the management condition of business enterprise more well and strengthen their investment confidence, we must carry on credit rating to listed companies, because enhancing valid supervision of listed companies is beneficial to the virtuous cycle of the whole market.Since China adopts the open police, small business enterprises and private enterprises mushroomed in quantities. However, in recent years, the financial state and conducting condition of business enterprise in our country fluctuated a lot as well as their manage method's, and a lot of medium to small business enterprises have been bankrupted, while some others are getting close to a collapse. This would make it difficult for investors to manage investment opportunity. Especially in 2001, with heavy decline in the stock market, investors lost confidence on stock investment more ever than before. Therefore, analyzing the management quality and investment values of a company by using scientific methods seems to be particularly important. And carrying on a to companies by adopting credit rating from institute is highly needed, which firstly would supply investors with investment references, then provide a loan references for bank and other creditors and also offer some references for stock department from all levels when evaluating the quality of listed companies.Compared with western developed countries, China has a long way to go in the evaluation of the amount of credit risk and management research. This difference is highlighted on management of credit risk and credit rating methods. Developed western have widely adopted to modern credit risk evaluation model based on option theories and the VAR method, and substantial default rate, breaking contract relativity and loss rate data are fundamentals of these models. Nevertheless, the market economy system in our country has just been built, the credit management is still not mature, and the credit rating industry also was on a stage of fast development, our country has not developed database resources used in constructing modern credit. So data problem has become the biggest obstacle in developing credit measures and credit crisis evaluation methods of our country.The credit rating industry of our country originated from the end of 80's in 20 centuries, and is still on a level of development and perfection. And the credit rating system usually adopts "beat the method of cent", Compared with the development of international reputation rating industry, we still have a long distance. And the financial market and credit system construction of our country begins very late, the credit risk management technique is relatively at a low level , relevant research on credit risk evaluation has jet not reached any particular accomplishment, and we didn't have substantial data for modern credit rating model construction. However, thanks to the fast development of computer science and soft ware industry, the author conducted a credit rating prediction on listed companies by using newly developed LVQ neural network method, aiming at uncovering the credit crisis evaluation, making a discussion on rating methods of listed companies, and attempting to exceed traditional credit model.In selecting indices, the author sorted financial state, capital quality state, loan crisis state and development potentiality as a whole. When conducting basic financial indices, capital size and industry differences are considered. The selection of those indices followed the landscape transverse and longitudinal comparison of indices. From the result, the LVQ neural network model is just approved, which can be well used for mode classification and credit rating declaration.The creativity of this paper focuses on the application of the LVQ neural network in credit rating, which is a significant discussion on constructing mathematical mode of credit rating. And it is easy to build a mode classification model by using computer and software existed. Compared with traditional statistic mode, the LVQ neural network has many advantages for it has no related assumptions. It allows incorrect data input and self-relativity between data and has a self-modulation function. But LVQ neural network also has their demerit, which is his poor interpretation.However, the author is expected to attempt some deeper approaches in case analysis. The classification of stylebook expects more exactitude rate, the selection of data is a little bit subjective, and the mode designation should be more reasonable. All of which is the field that needs improvement. The conclusion of this paper is that LVQ is applicable in credit rating of listed company in China, which is more effective than traditional statistic methods. On our recent conditions, the application of LVQ neural network model in credit prediction has great significance as well.