Research On Credit Risk Of Internet Finance Listed Companies In China

With the continuous development of Internet technology and the unceasing improvement of e-commerce platform,the Internet and the traditional financial industry have been deeply integrated with the traditional financial industry and developed rapidly.The rapid development of Internet finance also exposes many problems and risks,which brings challenges to China's financial supervision.Therefore,the research on credit risk of Internet financial companies in China not only provides investors with better choices,but also has certain reference significance for our financial supervision.We choose 123 Internet financial listed companies in Shanghai and Shenzhen A-shares from 2013 to 2017 in this thesis.eliminating six Internet financial listed companies with data missing.According to Z-Score model,the remaining 117 companies are classified into three categories:good financial situation,unstable financial situation and worrying financial situation.Then,20 Internet financial listed companies are selected from these three categories as Sample.Fifty-eight financial indicators and four non-financial indicators were selected from the window financial terminal.Then through the pretreatment of these indicators and the Kruskal-Wallis test,23 financial indicators were finally screened out.Five common factors were extracted from the 23 financial indicators by factor analysis.Fifteen Internet financial listed companies were randomly selected from the three categories as training set samples and five as test set samples.Then,the three types of training set data and test set data are substituted into the three credit risk evaluation models of Internet financial listed companies:Logistic regression model,decision tree model and support vector machine model.Comparing the forecasting results of these three models,the support vector machine model has the best effect on forecasting the credit risk data of Internet financial listed companies.The forecasting accuracy of training set samples is 91.1%,the prediction accuracy of the test set samples is 73.3%.The next is decision tree model,and the least effective is Logistic regression model.Comparing the error rates the Y=-1,that is,the financial situation is the lowest.The error rate of the decision tree model which discriminates the Y=1 error rate,that is,the financial situation is the lowest.