Default Credit Risk Assessment Of P2P Network Lending Based On Lasso-XGBoost Model

With the rapid development of network technology,the connection between finance and Internet is more and more closer,and the global credit model has also changed greatly.The traditional financial institutions produce credit and develop into Internet financial credit,at the same time,the demand of credit platform for credit risk control is getting higher and higher.Therefore,it is very important for P2 P lending platform and investors to classify and forecast customer credit default by establishing credit risk classification model.Firstly,this paper introduces the related research on credit risk of P2 P network credit,expounds the theory of classification model which can be used for credit risk assessment,and chooses two single models of Logistic Regression,Support Vector Machine and two integrated models of Random Forest and XGBoost to classify and forecast credit risk.Secondly,in the empirical stage,this paper analyses and studies the personal credit data of Lending Club,a famous foreign P2 P lending platform,cleans the credit data of Lending Club platform,filters the pre-processed data by Lasso algorithm,and balances the data by undersampling method.Subsequently,four classified prediction and comparison models,Logistic Regression,Support Vector Machine,Random Forest and XGBoost,were established.Finally,the model is compared with other validation indicators such as recall rate,accuracy rate and ROC curve.The results show the integrated model has better prediction effect,and XGBoost model has the best performance.Lasso algorithm and XGBoost model are combined to propose a new credit risk classification model: Lasso-XGBoost.Lasso-XGBoost portfolio model is more suitable for credit risk assessment,which significantly improve the accuracy of risk classification and prediction.