The Research On FMCG Listed Companies’ Financial Distress Prediction

With the rapid growth of national economy, the domestic and international marketdemand for our consumer goods is growing rapidly, at the same time as the industry intothreshold is low, the market competition is very fierce. In the course of the enterprises’business operations, their market risks are also increasing; financial distress prediction hasbecome an important part of modern enterprises’ financial management. Accurateprediction of financial risk, discovering the problems timely and guiding enterprises to takeappropriate measures for avoiding the risk have both theoretical value and practicalsignificance.According to public company’s financial distress prediction, the paper analyzes thedefinition of public companies in financial distress and a variety of financial distressprediction methods based on scholars’ research at home and abroad, which is good forestablishing a more practical, rational, scientific financial distress prediction model. As thetraditional statistical methods’ forecasts are usually inaccurate restricted by sample size andthe normal distribution, this paper applies Least Squares Support Vector Machine(LS-SVM) to prediction. This article introduced the principle of Support Vector Machine, builta fast-moving consumer goods listed company financial distress prediction model based onLeast Squares Support Vector Machine (LS-SVM) and studied the data released in public.The prediction accuracy is much higher than other methods based on the Least SquaresSupport Vector Machine (LS-SVM).This paper used empirical research methods, performing tests of Normal Distributiontest, Significance test, and Multicollinearity test on the selected data in order that the datacan be applied to forecast model. After that, this established a Logit regression model andLeast Squares Support Vector Machine (LS-SVM) model. By comparison the predictionaccuracy are obtained. The results showed that both results back to the generation and theaccuracy of test results, the LS-SVM are obviously better than that of the Logit regression.