By the overall economic environment change fast, the enterprise whose possibility of financial crisis is increased year by year. Therefore, to build a forecasting model of financial crisis is the important subject currently.
Forecasting model used in many ways on, but the old model of forecasting model, often only a single feature selection methods, with classification, rarely use the results of multiple classification models were compared with each other, so this study uses rough sets theory of feature selection, and using traditional method and data mining methods to classify and compare the accuracy between the forecasting model. To build a useful two-step financial crisis forecasting model.
This study used the rough sets to select financial variables, and classified, the re-sults showed that data mining methods analysis financial crisis index, that the main fea-tures of the financial crisis, The empirical results interpret the application of data mining methods in two stages of financial crisis forecasting model can actually reduce the mis-judgment of the traditional model, therefore, both in academic research and empirical work, will be contributed to the information users.
