Research on customer demand cluster method based on "graph"

Customer demand cluster analysis is an important basis of customer classification in management decision. With the development of structural models on customer's demands, the application of traditional cluster method based on demand characteristic vector is limited. In this paper, according to customer demand logic, we drew lessons from "Concept Graph" to define the formal structure of demand "graph". Based on this view, we proposed a model of Demand Structure Graph Cluster Pattern(DSG-CP), including: Similarity judgment and distance algorithm on different demand structural "graphs"; Clustering center selection of demand "graph"; Improved clustering algorithm and implementary system process. Finally, we verified the algorithm with customers' data of mobile communications in a certain region.