The Research Of P2P Lending Default Prediction On Model Based On Data Mining Technology

In recent years,the Internet and mobile communication technologies have improved all aspects of our lives.One of the most prominent is the development of financing,information intermediaries and payments based on the Internet,which laid the foundation for the rise of Internet finance.Internet lending is one of the Internet's financial services,which means that the process of borrowing money and signing contracts is completed through the Internet platform.However,compared to the risk control system of traditional finance,the risk control system of Internet finance has many problems in terms of standardization,accuracy and legality.P2P lending is facing the credit risk,market risk,operation risk and policy risk and other risks.Credit risk is the most important risk faced by P2P lending,and most of the P2P network platform for the present has not yet set up their own credit risk evaluation model.This will directly threaten the safety and survival of the P2P network platform.In order to realize the real-time forecasting of the default customers and improve the transaction efficiency of the P2P network platform,based on the traditional credit evaluation index system,this study used 55596 users information from the 360 financial network lending platform,these information included browsing records,bank water records,credit card billing records and other real-time information,and then extracted 545 features.Based on the extracted features,this study constructed P2P lending default prediction model using random forest,XGboost and boosting tree algorithms Then the three models were optimized respectively,and evaluated by the evaluation index of AUC,Recall,Precision.The results show that the XGboost model has the best predictive performance,which can provide some reference for the real-time prediction of default customers on P2P network in China.