Research on Customer Segmentation Based on a Two-Stage SOM Clustering Algorithm

Correct customer segmentation is the first step of effective CRM, not only play a role to optimize enterprise´s resources distribution or reduce cost, but also obtain more profitable market penetration. This paper proposed a two-stage clustering algorithm based on Self-Organizing feature Map, which avails of Self-Organizing feature Map to cluster the raw data initially, and then makes use of K-means method to merge the clusters resulted from the first step. Thus, the final clustering result is obtained. According to RFM method and constituents of customer value, customer segmentation indexes are selected. Based on the transaction database of one stock exchange in Shanghai, customer segmentation models are constructed and then Clementine 11.1 is used to mine the two. Afterward, the segmentation results are found and corresponding marketing strategy toward each cluster is constituted.