Research On The Application Of Data Mining In P2P Personal Credit Model

In recent years,China's economic developed rapid,people's income level has been significantly improved.In terms of consumer awareness people from the past can not afford to borrow in order to be willing to use social lending resources to carry out the transfer of personal funds.This also led to a variety of personal consumer loans,including personal travel loans,personal comprehensive consumer loans,personal loans,personal housing loans,personal car loans and other business has developed.This led to the emergence of a lot of P2 P platform,the scale of personal consumption credit is increasing year by year,the platform for personal credit risk control needs are also increasingly urgent.However,due to the lack of personal credit data,imperfect technology,lack of technical personnel and other issues,the evaluation of personal credit risk model is not able to meet the needs of credit institutions.The rapid growth of personal credit business and the imperfect credit risk management technology and formed a strong contrast,how to set up personal credit growth and match the risk management system model is currently China's credit industry priority among priorities.With the help of logistic model and data mining technology,based on making full use of the practice company nearly 10000 cases of personal credit data,the borrower will first through the preliminary classification of cluster analysis,combined with the actual situation and the historical experience of business borrowers will be divided into three or four categories.Then,according to the classified categories,the borrowing times and the borrowing amount are taken as the decision factors,and the decision tree is used to compare,and finally the better division method is obtained.According to the optimal division results,the borrowers are divided into three types: General borrowers,growth borrowers and risk borrowers.Then,the Logistic regression model was established by taking the borrower's age,educational background and average monthly income as independent variables,and the overdue condition as dependent variable.Adopt WOE method and virtual variable substitution of Logistic model was established,and compare the two models,in order to determine the borrower overdue probability,personal credit default risk assessment P2 P platform and forecast analysis of personal credit default probability.Based on data mining technology,this paper takes platform personal credit default status as the object of study,establishes the credit risk assessment model,and puts forward some suggestions for improvement on the risk management of P2 P platform.Based on the above analysis,we put forward the corresponding management measures of risk control: 1.in the loan review stage shall establish audit information system more reasonable,for the lender some important information indicators include monthly income,non real estate information audit should be strictly introduced;2.reasonable guide platform lender behavior,making certain incentive measures to encourage borrowers to consciously improve their own reputation;3.for cooperation with commercial banks or the P2 P credit reference platform,enrich the database,resource sharing,the establishment of a more reasonable credit system to achieve a win-win situation;4.to set up a risk sharing mechanism,in cooperation with the relevant insurance company,to avoid all risks only investors bear.