基于改进k-means聚类分析法算法的电商网站用户行为特征建模与分析  

Modeling and analysis of user behavior characteristics on e-commerce websites based on improved k-means clustering analysis algorithm

作　　者：葛青龙 GE Qinglong(Taizhou Vocational and Technical College,Taizhou Zhejiang 318000,China)

机构地区：[1]台州职业技术学院,浙江台州318000

出　　处：《自动化与仪器仪表》2024年第12期262-266,共5页Automation & Instrumentation

基　　金：浙江省中华职业教育科研项目立项《数字技术赋能高职院校精准就业供给服务路径研究》(ZJCV2023C12);台州职业技术学院校级“十四五”(2021年)规划教材《网店运营与管理》;2022年校级在线精品课程:台州职业技术学院2022年校级在线精品课程《网店运营与管理》。

摘　　要：电商网站中积累着海量的用户行为数据,面对其挖掘分析问题,研究基于Spark平台,选择k-means算法,对用户客服满意度进行聚类,结合孤独森林算法,进行算法优化,并将聚类结果输入到基于用户的协同过滤算法中。结果显示,相较于k-means算法等算法,改进k-means算法能获得更低的平均误差、更少的运行时间。当实验次数为10次时,k-means算法的平均误差为1.80,而改进k-means算法的平均误差为0.45。改进k-means算法的平均运行时间比k-means算法少2062 s。相较于基于用户的协同过滤算法,研究方法的准确率和F1值更大,其分别提高了8.39%和0.0479。此外,研究方法的运行时间更短,其平均运行时间为12.67 s,比基于用户的协同过滤算法少5.72 s。充分利用电商用户客服满意度,有利于提高推荐效果。E-commerce websites accumulate a massive amount of user behavior data.Faced with the problem of mining and analyzing it,this study uses the Spark platform,selects the k-means algorithm,clusters user customer service satisfaction,combines the Lonely Forest algorithm,optimizes the algorithm,and inputs the clustering results into a user based collaborative filtering algorithm.The results show that compared to algorithms such as k-means,the improved k-means algorithm can achieve lower average error and less running time.When the number of experiments is 10,the average error of the k-means algorithm is 1.80,while the average error of the improved k-means algorithm is 0.45.The average running time of the improved k-means algorithm is 2062 seconds less than that of the k-means algorithm.Compared to user based collaborative filtering algorithms,the accuracy and F1 value of the research method are higher,with improvements of 8.39%and 0.0479%,respectively.In addition,the running time of the research method is shorter,with an average running time of 12.67 seconds,which is 5.72 seconds less than the user based collaborative filtering algorithm.Fully utilizing customer service satisfaction of e-commerce users is beneficial for improving recommendation effectiveness.

关 键 词：K-MEANS算法 电商网站 用户行为特征 用户客服满意度 聚类 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]