Personal Credit Scoring Model Based On Internet Financial Credit Loan

Loan business has always been the main development business of banks.With the rapid development and popularization of the Internet,Internet finance is booming and personal credit loan business has gradually emerged.It has the advantages of low application threshold and fast processing speed,but banks have to bear more risks.So the good risk control ability has become the key to the success of banks.Therefore,how to establish a high-quality personal credit scoring model has become a hot issue in the field of Internet finance.Based on the personal reputation data of a bank,this paper establishes four types of personal credit score binary classification models,namely Logistic regression model,Lasso-Logistic regression model,Adaptive Lasso-Logistic regression model and ArctanLasso-Logistic regression model.First,the original data is encoded and descriptive statistical analysis.It is found that there may be some abnormal points and the multicollinearity problem may exist.Then make further statistical diagnosis and processing.Next,use the above four models for data analysis.Use gradient descent to estimate model parameters.Finally,two types of errors,ROC curve and AUC value are used to comprehensively evaluate the model prediction results.The empirical analysis results can be seen that the three penalized Logistic regression model--Lasso-Logistic regression model,Adaptive Lasso-Logistic regression model and ArctanLasso-Logistic regression model have higher prediction accuracy than Logistic regression model.Among them,the ArctanLasso-Logistic regression model has the lowest misjudgment rate,that is,the two types of high-quality customers and risky customers have the lowest probability of predicting errors.By comparing the ROC curve and AUC value,it is found that the model evaluation of ArctanLasso-Logistic regression model is also optimal.Therefore,the establishment of the ArctanLasso-Logistic regression model based on the personal credit scoring model of Internet financial credit loans is the best model.