Research on Personal Credit Evaluation of Commercial Banks Under Ensemble Learning Framework

In the light of the problems of the traditional personal credit evaluation index system and model, such as less evaluation indexes and small coverage, an index system based on a big data platform is established with a personal credit evaluation model on the basis of XGBoost (Extreme Gradient Boosting) proposed. By introducing Bayesian Optimization Algorithm(BOA), WOE and IV indicators to optimize the model, we try to improve the classification performance and generalization ability of the model, and finally verify the availability of the model through an interpretability analysis on the model by using SHAP(Shapley Additive expressions). Then through empirical analysis on the data from the Lending Club credit data set, the XGBoost-based model we constructed is used to compared with other classifiers in common use, which indicating that the model constructed therein delivers a good performance in various evaluation indices, including an AUC index of 0.73 and a KS value of 0.35. This can help financial institutions scientifically and quantitatively analyze the personal credit status of their customers, efficiently making loan approvals so as to avoid potential credit risks.