基于区间值数据动态聚类算法的客户市场细分

K均值算法(K-means)目前较为成功地应用于客户市场细分,但随着市场规模的扩大,面临着对于初始类个数敏感,易陷入局部极小值的严重问题,制约了聚类效果.提出基于区间值数据,以自适应欧氏距离作为度量的动态聚类方法,将客户的多维属性和基因算法结合提高类初始化质量,自适应地调整聚类数,并通过实验测试表现出较好的性能.

CLIENT MARKET SEGMENTATION BASED ON THE DYNAMIC CLUSTERING ALGORITHM OF INTERVAL DATA

Although being successfully applied to the field of client market segmentation currently,as a result of the enlargement of market scale,the algorithm of K-means is confronting the challenges from clustering initialization number as well as local minimum,and thus,the clustering effect is restricted.By combing the multidimensional property of client with genetic algorithm,A dynamic clustering algorithm based on interval data taking adaptive Euclidean distance as measurement is presented.The algorithm can improve the quality of clustering initialization and adjust the number of clustering adaptively,which has been proved to be effective by tests.