现代信息科技2024，Vol.8Issue(9)：126-130,5.DOI:10.19850/j.cnki.2096-4706.2024.09.026
基于贝叶斯优化-XGBoost的电商用户流失预测模型
E-commerce Customer Churn Prediction Model Based on Bayesian Optimization-XGBoost
摘要
Abstract
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.关键词
用户流失预测/贝叶斯优化/高斯过程/XGBoost/机器学习Key words
customer churn prediction/Bayesian Optimization/Gaussian process/XGBoost/Machine Learning分类
信息技术与安全科学引用本文复制引用
李宏明,庄伟卿..基于贝叶斯优化-XGBoost的电商用户流失预测模型[J].现代信息科技,2024,8(9):126-130,5.基金项目
福建省科协科技创新智库课题研究项目(FJKX-2022XKB023) （FJKX-2022XKB023）
