The Study Of Financial Risk Warining For Manufacturing Listing Companies Based On BP-Adaboost And Logistics Model

Along with the rapid development of China’s economy and the further opening ofthe capital market, China financial market has become the main financing platform ofthousands companies which was listed on the board, along with increasingly fiercecompetition of financial markets and superposition of various potential risks offinancial markets, every year on Shenzhen stock market and Shanghai stock marketthere are some part of companies being punished by CSRC for special treatment orstop listed processing with their own financial situation or other abnormal condition,so how to measure the financial risk of listed company accurately and timely have animportant practical significance, In this article we through to learn from the advancedpre-warning evaluation technology and method of developed capital market systemand financial market, establish a listed company financial crisis warning system notonly suitable for China’s stock market development but also have a good predictioneffect of financial crisis,thus,this system also has some theoretical significance ofChina’s listed companies’ financial risk’s measurement and early warningmechanisms.At home and abroad, there were some scholars study from artificial neuralnetwork model and Logistic regression model to establish risk and financial warningsystem which has achieved a good research result, However, the existing neuralnetwork model’s prediction has a certain instability, In order to solve this problem thestudy departure from the artificial neural network,Be combined with Adaboostalgorithm which can improved any weak learner to stronger learner and artificialneural network technology to build an early warning model to study China’s listedcompanies’ early financial warning status.The BP_Adaboost model in the article isessentially make the BP neural network as a weak classifier,by repeatedly training theBP neural network,forecast sample output,Combined with Adaboost algorithm to get astrong classifier which were consisted of the weak neural network classifier,finallyuse this classifier to take early warning analysis and research.In this paper’s study we select all key financial data and key financial indicatorof listed companies from CCER database, and finality select9index variables throughthe significant analysis and factor analysis, chose the94manufacture listed companywhic were punished by ST processing with financial exception and94normalcompany in the same industry, to collect its2008and2009’s financial data and get atotal of196sample space,and use Adaboost algorithm to build the BP neural network model and logistic regression model to take early warning analysis and study withthese financial data,the research process and the results show that BP_Adaboostmodel is feasible in the listed companies’ financial warning study and the predictioneffect is significantly better than the logistic regression model,Thus this model has agood prospect of research and application in the practical application.Finally,according to the problems found in the study and combine with the conclusion of themodel,put forward some policy suggestions to the problems of listed manufacturingcompanies.