Ant Clustering Algorithm And The Application In Telecom Customer Segmentation

With the trend of economy globalization and global economy informationization, it becomes necessary to increase information technology application in social economy for economic and social development promotion. Since 1980's, artificial intelligence has been put into practical application, and a new and commerce-directed subject----data mining has been put forward.Data mining technology has long been used for sales prediction and analysis in retailing, customer credit analysis and customer cheating analysis in financing, customer value analysis and sales prediction in telecommunications, etc. With the increasingly heated and complicated competition environment, the domestic enterprises are gradually informationized, too. In telecommunications, the telecom operators have or are going to set up an enterprise operation analysis system based on data warehouse by means of online analysis process and data mining to show situation of enterprise operation and marketing through intelligently analyzing the large amount of history data accumulated in daily business operation system so as to help management stuff make correct operation decision and then increase the marketing competitive strength of the enterprise.One of the basic ways to mine the data in the thesis is to make a comprehensive comparison to the clustering analysis, and to subdivide telecom customers by applying the improved clustering analysis so as to identify customer base with similar features, and to be the base of customer analysis and market forming tactics so that we can provide appropriate service at the right time for the right customers in the right way to meet their needs. This thesis mainly describes the application of group intelligent ant algorithm in clustering analysis, based on which we anlalyze the typical revised algorithm, and come to the conclusion that the AAC algorithm which is based on ant model is excellent in cluster speed and quality by comparing different ant action models.Considering the fact that most of the present clustering analysis works well with few dimentions and data, but not with multi- dimensional clustering, inspired by AAC algorithm, we propose a practical and efficient concurrent clustering algorithm by applying ant movement model in order to cope with the large amount of data and dimension in telecommunications, which is called hybrid ant clustering for it combination of hierarchy and density clustering. HAC algorithm partitions the large amount of data and then clusters and fuses them.Successfully applying the combined clustering algorithm to segment telecom customers. Analyzing the data of the customer information, communication action, service action, etc, and the relationship between communication action features of customer segments and service types, and between customer segments and profit, the experiment result proves that the clustering algorithm is efficient...