Abstract: A model was built to evaluate the customer value in terms of current value and potential value. This model used the Analytic Network Process (ANP) for weighting which considered the interrelationship among indexes, then calculated the customer value based on the weight and score of the indexes and then classified the customers by K-means. Taking a tobacco company for example at the end of this paper, qualitative and quantitative method was used to establish a customer value evaluation index system, ANP was used to weight indexes and classify the customers by K-means based on the evaluation result, and the marketing strategy of each customer group was analyzed at last. The proposed model can evaluate and classify the customer value more comprehensively and objectively.

Key words: customer value, classification model, Analytic Network Process (ANP), weight, K-means clustering

摘要： 建立两维度的客户价值评价模型，从客户当前价值和潜在价值两个方面评价客户价值。该模型首先运用网络层次分析法（ANP）对指标赋权，兼顾了指标间的相互影响，根据各指标权重和得分计算客户价值；然后使用K-means聚类算法对客户群进行细分，确定客户等级划分的标准，划分客户类别。最后以某市烟草公司零售终端客户价值分类为应用实例，运用定性和定量的方法建立多角度的客户价值评价指标体系，采用ANP确定指标权重，并根据客户价值评价结果使用K-means聚类进行客户分类，同时对每个类别的客户特征及其相应的营销策略进行分析，所提模型能够更全面、客观地对客户价值进行评价与分类

关键词: 客户价值, 分类模型, 网络层次分析法, 权重, K-means聚类