| With the development of China, enterprise faces more and more completion. It is necessary to establish model of financial crisis prediction to find out the potential risks to ensure the normal operation of enterprise. The current domestic and international financial risk pre-warning model has a lot of kinds, multivariate linear regression model, Logistic regression model, the technique of artificial neural network model and so on. By comparison we found that Logistic regression model is most suitable for listed companies in the financial data distribution characteristics.With the sign whether the financial status of the listed companies that is "special treated" is abnormal or not, the thesis selected the 104 companies of ST,*ST class company of 2009 as the research samples. Meanwhile 104 company of non ST are selected as the compare samples. The total number of the sample company reaches to 208.The time that financial crisis happened is taken as the T point. The three years' data of the sample company is used to be studied with SPSS17.0. It proved that financial ratios of the listed companies in China are not identical to the normal distribution by studying the characteristics of the financial ratios of the sample company. Logistic regression analysis method is chosen to establish the financial crisis forecast model. Secondly the study tests the 23 financial ratio significant differences between the two groups of last three years'. Then 9 ratios were selected as dependent variable with principal component analysis method to establish the model of, logistic regression. Last but not least, the model was used to test the 208 sample companies mentioned before. The rate of accuracy of the model for the last three years is perspective 84.62%,80.29%,71.63%. In addition, the paper establishes other three compare models by widen the sample companies from depth and width. The model is the most accurate one from the comparison of the results. The model is help for the regulation of listed companies, the evaluation of bank for customer credit standing, the regulation of enterprise itself. |