Research On Financial Early Warning Model In China's Listed Companies Based On Support Vector Machine

China capital market is changing the model gradually with the development of china economic innovation. Function of capital market is more important. At the same time those listed companies with bad operation get into financial crisis and the phenomenon appear gradually. Target of this article is set up financial prediction system. The financial prediction system can be used to prevent financial crisis.At first, we introduce the background of this research. And this thesis, analyses and contrasts the internal and foreign pre-warning model of financial data. Then we put forward the SVM model and advantages in solving listed companies financial pre-warning problem.Secondly, we introduce the basic theory of financial pre-warning system. Including the definition, connotation and character of financial pre-warning system and other main contents.Thirdly, basic mathematic theory and algorithm are provided, including statistic theory and support vector machine.At last, we build a model of listed company financial pre-warning by SVM and use the data of listed company to implement demonstration research. The main part of this study includes choices of index samples, model building, influences to this model under different model parameters, kernel functions and algorithms, and advantages compared to BPN model. We firstly apply SVM multi-classification method to financial multi-pre-warning problem, then extend simple two-class mode to three-class pre-warning mode, achieve better conclusion.The innovations of this thesis are that we establish the SVM financial pre-warning model achieved better results based on the traditional financial pre-warning model by fraction, then apply SVM multi-classification method which achieved good results of financial multi-pre-warning problem.