The Study Of Financial Distress Prediction Of Listed Companies Based On The Combination Forecasting Techniques

As of this year, China’s capital market has gone through the course of the lasttwenty-three years, enterprises are facing increasingly fierce competition, the slightestmistake business may fall into a financial crisis, the crisis of listed companies because ofabnormal financial positionis not uncommon. Therefore, the financial data of listedcompanies, listed companies’ financial crisis early warning model to reveal the risks, hasbecome the common concerns of managers of listed companies, investors and creditors andother stakeholders.This paper selected financial data of146listed companies from multiple industries,including73well-listed company’s financial data, and paired73special treatment (ST said)the company’s financial data, divided into49pairs of training samples and24of the testsample. Research scholars warning of enterprise financial crisis, asked the20financialindicators to build a financial early warning indicator system. T-test and non-parametrictest will be more original predictor filter indicator system, making predictions is morereasonable, and the use of factor analysis to eliminate collinearity between indicators, theseven factors. Were used multiple discriminant analysis, logistic regression analysis andsupport vector machine classification to determine the training sample data to constructthree single financial crisis early warning model; constructed on this basis, a linearcombination of the prediction model and a neural network combined forecasting model,and finally the use of the test sample of five model effective verification on five test resultswere compared. The research results show that the combination forecasting model basedon neural network methods predict the effect of corporate financial distress, compared witha single method, the prediction accuracy is not significantly increased, but the model ismore stability.