动静态属性数据相结合的客户分类方法研究

研究了一种客户动态、静态属性数据相结合的客户分类方法。提出了客户时间序列的加权处理方法,并应用客户时间序列的统计特征作为聚类特征向量,采用混合式遗传算法对客户聚类,使每一类客户具有相似的时序特征。在此基础上将聚类结果与客户的静态属性数据相结合,对客户进一步分类。实验结果表明,与传统的基于静态属性数据的客户分类方法相比,本文的方法提高了客户分类的准确性。

Customer Segmentation Based on Integration of Dynamic and Static Attributes

A novel customer segmentation method based on integration of customer’s static and dynamic attributes is developed.A weighted method of customer’s time series is proposed and statistical features of time series are adopted for customer clustering,which make each group of customers have similar sequence feature.A genetic algorithm approach is adopted to improve the clustering quality.Clustering result is combined with the static attributes of customer for further segmentation.Experimental result indicates that compared with traditional segmentation methods,the proposed method improves the accuracy of segmentation.