Clustering Fusion Algorithm And Its Application In The Telecom Consumer Segments

Clustering analysis has been one of the most important research direction in the field of data mining and knowledge discovery,clustering technology has also been more and more popular, and is widely used in many areas,such as engineering, biological medicine, marketing, business intelligence and decision analysis. Although the development of clustering analysis technology has become more sophisticated, also produced many great algorithms such as K-Means, BIRCH, DBSCAN, SOM, the requirements for clustering algorithms have become increasingly stringent. Each clustering algorithm has some limitations, and not a single clustering algorithm can be suitable for all situations. Therefore, an idear to integrate the advantages of different clustering algorithms to get better results was first proposed in 2002 and quickly became the research focus, this is the clustering ensemble. Since clustering ensemble is an emerging technology, itself still in the exploration and development stage,, and there are still many problems that require further improvement.The major work in this paper is to present a weighted PCA-based cluster ensemble approach,which produces cluster members via a deterministic initializing cluster means.Meanwhile, this approach sets weight to all cluster members according to the quality coefficient which is the evaluation of the clusters'qualities, and the K coefficient which is the evaluation of the influence the number of K contributes to the clusters. Finally, the improved clustering ensemble algorithm was used, and the data mining of a certain city's telecommunications company was used as the background, aims at customer calls,messages and other attributes of the customers'action to improve its customer segmentation.