The Empirical Research On Credit Risk Evaluation Model Of Online Peer-to-peer Lending New Borrower Based On Support Vector Machine

The P2 P Lending has rapidly rising in the world after the first P2 P platform established in the United Kingdom since 2005.China's first P2 P platform "PPDAI " was established in 2007,and now it's 2448 P2 P platform.Behind the exponential increase of P2 P platform,there are various problems,withdrawal difficulty,platform collapse,reconnaissance,absconding with money and so on According to the data from ‘WDZJ.COM',by the end of December 2016,the number of problem platforms increased from almost zero in 2010 to 1741 at the end of December 2016.All of this causes the social concern as well as the academic community's research interests.For the P2 P platform itself,the most serious is individual credit risk,which determines whether the platform can continue to survive.Therefore,this paper mainly carries on the empirical research to the individual credit risk of the borrower of P2 P platform and constructs a reasonable personal credit risk assessment index system,so as to provide a more rigorous,cost-effective and low-cost credit audit method for the platform.Based on the large number of relevant literatures and related materials,this paper analyzes the current situation of domestic and foreign scholars and the development of P2 P industry in China.Based on the previous studies,this paper chooses the new borrowers from "RenRenDai" platform as the object of this article,the research of credit risk assessment index system in the new borrower is formed the twenty-five borrower evaluate indicators and build the improved P2 P network loan platform new borrower credit risk assessment index system are on the basis of China's traditional commercial banks individual credit risk assessment index system and in accordance with the development of China's P2 P industry and fully considering the credit environment of the " RenRenDai " and the content of new the borrower credit information.And finally using MATLAB and IBMSPSS software with the same sample data,the support vector machine method and the Logit regression method were used to analyze the index system before and after the improvement,and the accuracy of the classification prediction results was compared and analyzed.The empirical results show that the improvement of the credit risk assessment index system(98.93%)of the new P2 P lending platforms in this paper has better prediction accuracy than the pre-improvement index system(94.65%).For new borrowers default rate(0.28)forecast is better than that of the selected platform sample set itself the default rate(0.19),the borrower's credit risk prediction is more precise,thus it can be reduce the actual incidence of the default borrowing from the sources;The results of the evaluation of the index system before and after the improvement by the support vector machine method are better than those obtained by the Logit regression model(89.3% and 90.9%).The results show that the index system is constructed in this paper and the use of support vector machine(SVM)method in the P2 P network platform for new loan borrower credit risk evaluation has certain practicability and superiority of evaluation results,perfecting the credit evaluation system to P2 P industry has a certain practical significance.