基于RFM模型的电子商务顾客细分研究  

Customer Segmentation of E-Commerce Based on RFM Model

作　　者：吴涛[1,2] Wu Tao(University of Science and Technology of China,Hefei Anhui 230026,China;Anhui Industry Polytechnie,Tongling Anbui 244061,China)

机构地区：[1]中国科学技术大学,安徽合肥230026 [2]安徽工业职业技术学院,安徽铜陵244061

出　　处：《铜陵学院学报》2020年第5期55-59,共5页Journal of Tongling University

基　　金：安徽高校人文社科研究重点项目“中小城市电子商务人才需求研究——以铜陵市为例”(SK2017A0925);“'一带一路’背景下中资矿业开发与南美文化的冲突与整合——以厄瓜多尔ESCA公司为例”(SK2018A0973);安徽省高校优秀人才支持计划项目(gxyq2020276)。

摘　　要：;电子商务蓬勃发展的同时,市场竞争也日趋激烈。为提高电子商务企业的核心竞争力,需要对企业顾客进行精准细分,进而采取个性化的营销策略,提升顾客满意度与忠诚度。文章建立RFM模型,分别使用RFM分析、K-means、K-means++三种方法对顾客最后购买日期与当前日期的间隔、顾客在某时段内的购买次数、顾客在某时段内的消费总金额3个顾客行为指标进行分类,并对三种方法进行了评估。通过对比发现,K-means++算法的指标分类结果最为合理。最后将K-means++聚类结果进行视化,使企业人员可以准确直观地对顾客价值进行识别,进而针对不同价值的顾客群体进行差异化管理。With the vigorous development of electronic commerce,the market competition is becoming increasingly fierce.In order to improve the core competitiveness of e-commerce enterprises,it is necessary to accurately subdivide enterprise customers,and then adopt personalized marketing strategies to enhance customer satisfaction and loyalty.In this paper,SPSS software is used to sort out the online sales data of an e-commerce company and establish an RFM model.On this model,customers are subdivided by three methods-RFM analysis,K-means++clustering analysis and K-means clustering analysis,and the three methods are evaluated.Among them,the K-means++algorithm with the number of elustering centers is reasonably selected by the contour coefficient method,which makes up for the deficiency of the traditional K-means algorithm in e-commerce customer segmentation and is more scientific and credible than RFM analysis.Finally,through the data visualization of K-means++elustering resuls,enterprise personnel can accurately and inti-tively identify customer values and carry out differentiated management for customer groups with different values.

关 键 词：电子商务 顾客细分 RFM模型 聚类分析 K-means++ 

分 类 号：F713.365[经济管理—产业经济]