林　勤　薛　云　杨柏高

1(广东医学院信息工程学院　广东 东莞 523808)

2(华南师范大学物理与电信工程学院　广东 广州 510006)



基于变异系数的双聚类算法及其在电信客户细分的应用研究

林勤1薛云2杨柏高2

1(广东医学院信息工程学院广东 东莞 523808)

2(华南师范大学物理与电信工程学院广东 广州 510006)

摘要针对传统客户价值细分方法不够精细化的问题，提出一种基于变异系数的双聚类算法。该算法选用了变异系数作为相似性度量，运用启发式贪心策略，通过迭代增删行列的方式挖掘出客户消费记录中局部消费行为相似的客户群体。以某电信公司的电信客户细分为实例，将所提算法与K均值(K-means)算法进行性能比较，实验结果表明，所提算法具有更优的客户细分能力和更强的客户行为可解释能力。因此，它更有助于指导企业制定差异化营销策略。

关键词变异系数客户细分差异化营销双聚类K均值

ON VARIATION COEFFICIENT-BASED BICLUSTERING ALGORITHM AND ITS APPLICATION IN TELECOMMUNICATION CUSTOMER SEGMENTATION

Lin Qin1Xue Yun2Yang Baigao2

1(School of Information Engineering,Guangdong Medical College,Dongguan 523808,Guangdong,China)2(School of Physics and Telecommunication Engineering,South China Normal University,Guangzhou 510006,Guangdong,China)

AbstractTo improve the refinement degree of traditional customer value segmentation method, we proposed the variation coefficient-based biclustering algorithm. The algorithm selects the variation coefficient as the similarity measurement, applies the heuristic greedy strategy, and by the way of iterating the rows’ or columns’ insertion and deletion, the algorithm mines the customer groups with similar local consuming behaviours from their consumption records. Taking the telecommunication customers segmentation in a certain Telecom as the example, we compared the performances of the proposed algorithm with k-means clustering algorithm. Experimental result indicated that the proposed algorithm has better ability of customer segmentation and stronger interpretable ability on customer behaviours. Therefore, it is more conducive to guiding the enterprises to develop differentiated marketing strategies.

KeywordsVariation coefficientCustomer segmentationDifferentiated marketingBiclusteringK-means

0引言

近年来，随着电信运营商网络基础设施间差距日益缩小，所涵盖业务日趋同质，电信企业的发展模式逐步从以产品为中心的模式向以客户为中心的模式转变。精确识别和细分客户市场，有助于电信企业合理配置企业资源，提高资源利用效率，降低运营成本；同时也有助于发现不同类别的客户对服务的不同需求，为其提供个性化的服务，从而使运营商和客户价值发挥到最大，实现电信企业和客户之间的双赢。因此探究精细的数据挖掘方法对客户进行细分是目前数据挖掘应用的一个非常热门且具有重要应用价值的研究课题[1]。……