Clustering Fusion Algorithm And Its Application In Mobile Channel Management

Along with the increasingly rapid development of information society and increasing information data, database technology advances rapidly. To meet the need of information, researches of data mining methods are paid more and more attention, the cluster analysis as an important part of data mining study has become one of the hotspots. The direct purpose of cluster analysis is to provide information support for solving problems by dividing data set into valuable groups. However, the fact is that mass data have many kinds of data type, and have different distribution shapes. Based on this fact, more demands are proposed for cluster analysis. Judging from the previous domestic and international research literature, too many researches focus on single algorithm, however, facing more demands of practical application, single algorithm shows more and more problems, and clustering Fusion study has become the new hot spot. In addition, cluster fusion study focuses in merge all the cluster members, though researches of selection of cluster members are so shortage. Based on the current research situation, the paper is to study algorithm of selection of cluster members and weighting of cluster and fusion, and its application in mobile communications companies.After studying and analysis of existing clustering and clustering fusion algorithm, the paper proposes a selective weighted clustering fusion algorithm for its deficiency, which firstly selects high quality fusion members by calculating difference degree and accuracy of cluster members. The key point of calculating accuracy is to get reference. It is by using a fusion technology, which is only fusing two cluster members once, to generate fusion set as the reference. Then, by means of the method of attribute weighted to weigh fusion members, which transforms every fusion member into one attribute of the original data, it accomplishes the weight of fusion members by way of attribute weighting. Experimental results show that the new algorithm can effectively deal with differences in the quality of cluster members, get better clustering results compared to former cluster fusion. Moreover the new algorithm has good scalability. Finally, the thesis applies the selective weighted clustering fusion algorithm to business channel management for a mobile company, which gains helpful channel information for channel management and verifies effectiveness of the algorithm by experiment, by applying date mining method to consumer behavior, channel service data and sales data in channel, and by analyzing service ability of channel for channel set.