Study On P2P Lending Personal Credit Evaluation Based On The Logistic Model

P2P lending is a a new model of borrowing between micro,small,medium enterprises and person on online.As of December 31,2015,the P2 P lending platform number or loan deals throughout the year,more than 2010 full-year figures thousands of times,so rapid pace of development is bound to appear all sorts of problems.Mainly in the following four aspects: First,legal supervision and examination can't keep up with the speed of development;Second,reference way of platform is multifarious,there is no industry standard,resulting in credit risk;Third,because of the lending is carried out through online,the true use of lending to borrowers are not privy to the lender,lead to moral hazard;Fourth,according to the year 2015 of "Peer-to-Peer lending" lending data shows that demand more than supply in capital financing market,it still cannot satisfy the loan demand of most people.In view of the above four aspects,this paper holds that the credit risk of P2 P mode can objectively delve into the field.Through consulting literature,the existing five kinds of personal credit evaluation model,the Logistic regression model is the only thing can be predicted at the same time the lenders will whether fail bids and whether exist possibility of default.At the same time,for the platform,results can also according to the default customer information,establish a large database,thus reducing platform operating risk.Based on Logistic model,this paper constructed the Performance-Default Logistic Regression Model(PDLRM)and Failing-Full Bids Logistic Regression Model(FFBLRM)respectively.From the FFBLRM result,we can see that on the Logistic regression model in 18 variables that probability of full bids,gender,mortgages,car loans,loan amount,loan interest rate,credit reports,etc.that information of 11 items have significantly influence on borrowing of the success of the lenders.Can be seen from the PDLRM,in 22 variables affect the probability of default,the age,car loans,credit reports,etc.that information of 7 items have significantly influence on whether the borrower defaults.Through verification,the model prediction effect is obvious,that has a certain practical significance,in view of the problems at the same time,put forward policy suggestions.According to the results,the FFBLRM forecast was up to 97.9%,the PDLRM forecast was up to 98.5%.For P2 P network problems in the credit industry,four suggestions are put forward.