Research Characters Of Mobile Phone Subscribers Network And Market Segmentataion By Social Network Analysis

Mobile phone log provide detail of subscribers'communication pattern. It can be mapped to a series of social networks which describes the mobile telecomm market. We analysis the time cycle, satiability and evolution of these networks through social network and complex network theory. We create a'Phone Call Network'model, and select the mobile subscribers as the vertices and the phone call relationship as the edges, the phone call frequency, lasting time as the volumes of the edges. And then we implement the model by an eight month period real phone call record data from a branch company of one of the major telecom service provider of China. We find that the'phone call network' has a noticeable 7 days time period and a holiday effect. We make an ARMA model to analysis the time characters of the network. The network records the position of every phone call, with these information we paint a voronoi diagram of the space distribution of the network. By this diagram we learn that the space distribution of the'phone call network'is discrete, there are only very small number of'hot islands'. Comparing with the BA, ER random network model, we learn the degree distribution, small world character, etc of the network model. Then we make the segmentation of the market by social network community algorithm and lay it out in 3D motion.