A Model and Empirical Analysis on Financial Distress Forecasting of Listed Companies Based on Least-Square Support Vector Machine

This paper applies least-square support vector machine (LS-SVM), a statistic machine learning methods, and establishes a model of financial distress prediction. Based on information of listed companies in Shanghai and Shenzhen during the year 2005 to 2006, the paper gives an empirical analysis of financial distress prediction. Research conclusions show that prediction and self discriminate capability of the prediction model is increasing by year as the coming of financial distress, and the remained discriminate capability is higher than the prediction capability.