Financial Crisis Prediction Of Listed Companies Based On Rough Set Theory And Artificial Neural Network

Listed companies’financial crisis prediction has been a hot topic over economics, statistics and computing science. As this study could effectively predict the possibility of financial crisis, researchers, politicians and businessmen have all focus on this field.This paper related theory with practice and proposed a financial crisis prediction method based on rough set and artificial neural network according to the characteristics of companies’financial data. This method has an advantage of low time complexity and high prediction precision over other ones. The main research contents are as follows:Firstly, according to the characteristics of companies’financial data, attributes discretization is inevitable. Looking back on existing discretization methods of continuous attributes, the paper proposed the method based on dynamic neighbor domain clustering which effectively diminish complexity of data processing and make it more suitable for large amount and high dimension data analysis.Secondly, establish the listed companies’financial crisis prediction model using uncertain data analysis ability of rough set and error toleration and prediction ability of neural network. This model is quite precise and could judge whether the tested company will confront with financial distress problem.Thirdly, compare the earlier proposed method with ways of purely using rough set or neural network and effectively verify the effectiveness, time efficiency and precision of the RS-ANN based listed companies’financial crisis prediction method.Finally, the conclusion and next step research goal are given.