Under the background of the big data era,the logistic regression model commonly used by domestic commercial Banks has been struggling to cope with the existing huge and complex data sets.Therefore,it is necessary to find feature engineering methods that can improve the performance of the current model,and explore the effectiveness of other machine learning algorithms in the field of personal credit evaluation.Feature engineering is the key to machine learning modeling.In this paper,the XGBoost algorithm based on extreme gradient boosting is used to study the feature engineering in the credit evaluation model based on machine learning.This paper first studied the XGBoost algorithm deeply and rearranged its principle and structure.And then,based on the real personal credit data published on the Lending Club platform,this paper conducted the modeling,feature selection,and the empirical research,which shows that,compared with traditional methods,the XGBoost algorithm not only has obvious advantages in nonlinear feature identification but also has a better effect on the accuracy and differentiation of credit classification. |