基于选择性聚类集成的客户细分

发布时间：2018-01-13 11:03

本文关键词：基于选择性聚类集成的客户细分　出处：《计算机集成制造系统》2015年06期 　论文类型：期刊论文

【摘要】：针对数据密集型企业的客户细分问题,提出一种基于选择性聚类集成的客户细分框架。在聚类集体生成阶段,根据数据来源和业务需求构建统一的客户视图,将客户特征划分为若干子集后再分别对客户对象聚类,通过评价函数选择高质量的个体标记向量生成聚类集体;在聚类集成阶段,构建记录簇标记所覆盖的相同对象个数的重叠矩阵,利用重叠矩阵计算各簇权值,最后选择最具代表性的簇参与集成。通过某企业客户细分的实证研究表明,该框架可以有效识别出不同价值和消费行为习惯的客户群,为企业制定产品营销方案提供依据。
[Abstract]:To solve the customer segmentation problem of data-intensive enterprises, a customer segmentation framework based on selective clustering integration is proposed. In the stage of clustering collective generation, a unified customer view is constructed according to the data sources and business requirements. The customer features are divided into several subsets, and then the customer objects are clustered separately, and the high quality individual marker vectors are selected by the evaluation function to generate the clustering group. In the cluster integration phase, the overlapping matrix of the same number of objects covered by the record cluster tag is constructed, and the weight of each cluster is calculated by using the overlapping matrix. Finally, we choose the most representative cluster to participate in the integration. The empirical study of customer segmentation shows that the framework can effectively identify different customer groups with different values and consumer behavior habits. To provide the basis for the enterprise to formulate the product marketing program.
【作者单位】：温州大学建模与数据挖掘研究所;温州大学物理与电子信息工程学院;
【基金】：国家自然科学基金资助项目(61402336) 浙江省科技计划资助项目(2012C33086,2013C31138) 浙江省自然科学基金资助项目(LQ12F02008)~~
【分类号】：F274
【正文快照】： 0引言近年来,随着企业信息化水平的提高和数据存储技术的发展,不少数据密集型企业积累了海量的客户业务数据,如何对这些数据进行深层次的分析,从中获取客户的行为特征和消费偏好,并有针对性地开展营销与服务,挽留高价值客户,成为企业面临的重要问题[1-2]。客户细分方法[3-6]是

【参考文献】