The Research Of Forecasting Model For Financial Distress Of Chinese Listed Company

With the continuously development of our country capital market and, the continuously perfect of the listed company information publishes system there are more listed companies being special treatment. This phenomenon not only threats the company's subsistence itself but also brings investor and creditor huge great damage. Therefore it is very meaningful to set up a valid and practical financial crisis early-warning model.The main study process is introduced step by step as follows. Firstly, we used T examination and relativity analysis method to carry on preparing process. Secondly, we used rough sets as a preprocessor for reducing the financial ratios. Thirdly, taking BP nerve network and support vector machine as main forcast model to establish the composing financial crisis early-warning model. At last, for verifying the effectiveness of our model we used the traditional Fisher discretion as a contrast model to carry on the substantial evidence research again.The study result shows that the RS-BPNN and RS-SVM early-warning model have good forcast ability and, its total accuracy all above 97%. In addition, comparing with Fisher discretion which based on rough set the two composing model have a better performance in aspect of total early-warning accuracy and the first mistake occurrence rate.