基于机器学习组合模型的用户消费行为预测  

Prediction of User Consumption Behavior Based on Machine Learning Combination Model

作　　者：张峰[1] 张丽娜[1] 李静静[1] ZHANG Feng;ZHANG Li'na;LI Jingjing(Hebei University of Engineering,Handan056038,China)

机构地区：[1]河北工程大学,河北邯郸056038

出　　处：《智能物联技术》2022年第2期19-27,共9页Technology of Io T& AI

基　　金：河北省自然科学基金项目(A2021402008,F2021402010);河北省高等学校科学研究项目(ZD2020185,QN2020188)。

摘　　要：大数据背景下,如何利用大量的用户消费行为数据精准识别高质量的用户和渠道,进而预测用户的未来需求,已成为企业面临的难题,对用户消费行为进行深入挖掘与分析具有重要的现实意义。本文首先对用户消费行为数据进行可视化分析;其次,结合随机森林和Logistic模型,提出了一种基于机器学习组合模型的用户消费行为预测方法。结果表明:该方法的AUC值为94.85%,明显优于其他模型。该方法可为企业分析用户消费行为、制定科学的营销策略提供借鉴。In the context of big data,it has become a difficult problem for enterprises to accurately identify high-quality users and channels by using a large number of user consumption behavior data,so as to predict the future needs of users.Therefore,in-depth mining and analysis of user consumption behavior has important practical significance.Therefore,this paper first carries on the visualization analysis to the user consumption behavior data;Secondly,combined with random forest and Logistic,a predictive method of user consumption behavior based on machine learning combination model is proposed.The results show that the AUC value of this method is 94.85%,which is obviously better than other models.It can provide a good reference for enterprises to analyze consumer behavior and formulate reasonable strategies.

关 键 词：用户消费行为 组合模型 随机森林 LOGISTIC 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] F274[自动化与计算机技术—控制科学与工程]