Financial Warning Model Research Of Listed Companies Based On RBF Neural Network

In modern society, the companies are facing various risks; financial risk is aimportant one. The companies will face financial crisis if they can not take effectivemethods to deal with the financial risks. The financial environment of companies is goingthrough a series of changes, which requests a new and better financial risk management.The defense and control to the financial risks of the companies, is one of the hottest topicsin the financial theory research and the financial practice research. Empirical research onfinancial crisis warning system is one of the most important research subjects in manyforeign countries. There are a lot of financial crisis warning models which use of variousmethods. Each kind of model has the advantages and disadvantages, and the precision andspeed of model waits for improving. Based on the mature experiences abroad and home,and the character of capital market of China, this paper builds the system of financialratios used the public financial data of selected listed companies as samples, and applyRBF (radial basis function) neutral network to the model of financial warning. We hopebuild a better model through using of the advanced technology in the field of neutralnetwork.The study in this paper includes five main parts: firstly, we explain the backgroundof this study and the definition about financial crisis and financial warning. Secondly, wereview the theory of financial distress in foreign and domestic area, and describe the ideaand methods applied in this paper. Thirdly, we build the system of financial ratios throughselecting the sample companies and financial ratios. We select the "ST" companies in2005 and 2006, and normal companies according to the relevant industries and years. Thesample is new and the amount of it is large, it is good at the forecast precision of themodel and the guidance of practice. In the fourth part, we introduce the RBF artificialneural network, and build two models using BPNN and RBFNN respectively. The resultindicates that the forecast precision and speed based on RBFNN are better than that basedon BPNN, and the result is extremely ideal. The last part is conclusion, and some suggestfor guarding the financial crisis of listed companies.