E-commerce Customer Churn Prediction Model Based on Bayesian Optimization-XGBoost
In response to the problem of predicting E-commerce customer churn faced by E-commerce companies during their development process,an E-commerce customer churn prediction model BO-XGBoost is proposed,which combines Extreme Gradient Enhancement Regression Tree(XGBoost)and Bayesian Optimization method(BO).By comparing the model with the XGBoost model optimized by commonly used random search and grid search methods,it is verified that the proposed model has a higher F1 score and better efficiency.To further evaluate the prediction model,the BO-XGBoost model is compared with BO-LR,BO-SVM,BO-RF,and the unoptimized XGBoost model.The results show that the BO-XGBoost model perform the best in accuracy,accuracy,recall,and F1 score.At the same time,the recall rate,which is more important in the field of E-commerce customer churn prediction,reaches 95.26%,significantly leading other models,which indicates that the BO-XGBoost model has achieved good results in predicting E-commerce customer churn.
国家科技期刊平台
NSTL
万方数据
