基于拓展自组织神经网络方法的客户市场细分  

An Extending SOM Method for Customer Marketing Segmentations

作　　者：仲秋雁[1] 莫佳卉[2] 

机构地区：[1]大连理工大学管理学院,大连116023 [2]哈尔滨工业大学管理学院,哈尔滨154000

出　　处：《中国管理科学》2008年第S1期551-555,共5页Chinese Journal of Management Science

摘　　要：准确的客户市场细分对客户关系管理具有重要作用。现有的聚类算法由于自身的缺陷无法很好的应用在市场细分中。本文借鉴拓展自组织神经网络方法的思想,运用Ward's指标对SOM聚类结果进行合并,从而得到有效的客群。运用该方法在某零售企业的客户数据中探索出四类客群。ANOVA分析证明了该方法得到结果的可行性,并通过与K-means方法的比较证明了该方法的准确性。Accurate marketing segmentation is important for customer relationship management.Due to the drawbacks of the existing clustering algorithms,they cannot be applied to marketing segments.To get valid customer groups,this research adopted Ward's index to combine the results of SOM based on the idea of extending SOM. This method was applied to customer samples of one retailing company to achieve four customer groups.ANOVA test demonstrated the results of this method were valid,and the comparison results with K-means proved that this method was accurate.

关 键 词：客户市场细分 聚类分析 自组织神经网络 

分 类 号：F274[经济管理—企业管理] F224[经济管理—国民经济]