Research On Credit Risk Analysis Model Of Personal Loan Based On XGBoost

In recent years,with the rapid development of China’s national economy,people’s living standard is gradually improving,and the consumption concept has also begun to change.The number and scale of personal financial services,such as personal consumption loans,have increased significantly.Therefore,financial institutions want to substantial revenues from the credit loan business,but high returns are often accompanied by high risks.How to improve the risk management level of credit loans has become a key issue to ensure the healthy and sustainable development of financial institutions,and the current credit loan business has the characteristics of low single amount and many transactions,which determines that credit institutions need to build intelligent and efficient to manage risk.This paper establishes a credit risk assessment model for personal loans based on XGBoost algorithm,and explores the causes and rules of default through data mining and analysis of a large amount of historical credit data.This paper mainly studies two aspects.On the one hand,it constructs XGBoost credit evaluation model based on feature importance selecting features,and compares it with Decision Tree and Logistic Regression,and finds that XGBoost model has better classification effect.On the other hand,feature importance selecting based on the output of XGBoost model was compared with the features selected based on the information value,and it was found that feature engineering based on the feature importance of XGBoost model was more capable of selecting effective features.The above research shows that the big data credit risk control model based on XGBoost algorithm can improve the efficiency of credit risk management.