Application and research of clustering fusion algorithm in communication network prediction

作者：Li, Xiaolei^*

来源：Cluster Computing, 2019, 22(S4): S8429-S8436.

DOI：10.1007/s10586-018-1865-0

• Data mining
• Clustering
• Difference
• Consensus function
• Customer segmentation

In order to solve the problem of clustering fusion algorithm, such as key parameter setting, fusion of "soft" hard clusters, design and selection of consensus functions, we optimize the K-means algorithm. However, this method has many problems in practical application. It requires professionals to specify the number of clusters and make empirical judgments on the results. The improved algorithm of clustering fusion is introduced into the customer segmentation. Based on the data mining of the mobile phone business of a telecom company in a certain city, customer segmentation is carried out, according to the characteristics of customer calls, SMS and other attributes. The results show that the improved clustering fusion algorithm can effectively solve the above problems and get a reasonable clustering result. At the same time, by analyzing the CO association matrix, we can obtain each customer's belonging class. The purpose of dividing the results is achieved, which makes the data mining more intelligent.

• 出版日期2019-7
• 单位宁波大红鹰学院