基于效用分析的客户聚类方法研究  

Method to Customers Clustering Based on Utility Analysis

作　　者：王红军[1] 陈庆新[1] 陈新[1] 郑德涛[1] 

机构地区：[1]广东工业大学CIMS实验室,广东广州510090

出　　处：《计算机集成制造系统-CIMS》2003年第3期237-241,共5页

基　　金：国家自然科学基金资助项目(59985002);国家863/CIMS主题资助项目(2001AA411350);教育部教师重点项目基金。

摘　　要：企业实施批量定制的前提是分析客户的需求和偏好,正确地进行客户群的划分。综合应用响应曲面法、适应性关联分析技术和系统聚类法,提出了根据客户需求效用值进行客户群聚类划分的新方法;采用响应曲面法生成典型样品进行客户调查,通过适应性关联分析法分析调查结果,并计算客户需求的效用;采用系统聚类法对客户进行聚类划分,得到了具有不同需求效用的客户群;采用排序法确定不同样品的重要度,计算各样品在客户群中受欢迎的程度,优选出客户群偏爱的典型样品。The precondition for enterprises to carry out mass customization is that enterprises must analyze customers' needs and favorites, and partition customers into clusters. The proposed method to cluster customers is to combine response-surface method, adaptive-conjoint analysis technology and system-clustering method according to the value of customers needs utility. Customers' responses are collected from the questionnaires designed by using the response-surface method. Customers' utilities are analyzed and calculated by the adaptive-conjoint analysis procedure, and customers are partitioned into clusters that have different requirement utilities by the system-clustering method. The typical product of favoritism of customers clusters is evaluated and selected by sorting samples and calculating their scores.

关 键 词：适应性关联分析法 系统聚类法 响应曲面法 效用函数 批量定制 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TH122[自动化与计算机技术—计算机科学与技术]