基于通讯数据的社群聚类分析

Analysis of Community Clusters Based on Communication Data

该文根据现有部分人群的通讯数据,统计并分析所有用户的通话时间和通话次数.根据推导权值计算公式Wij=αcos(→ui,→uj)+βT^2ij/Ti*Tj N^2ij/Ni*Nj计算用户个体个性与用户通话时间和通话次数的加权相关性,构建出相应的权值矩阵,并分别用层次聚类法和K-Means聚类法对所有用户进行聚类,综合分析两种聚类方法,得出已知人群分四类最合适的结论,具体聚类结果见正文.

Based on the existing communication data of some people,this paper counts and analyzes the call time and call times of all users.According to the derivation weight calculation formula Wij=αcos(→ui,→uj)+βT^2ij/Ti*Tj N^2ij/Ni*Nj,the weighted correlation between the user's individual personality and the user's call time and call times is calculated,and a corresponding weight matrix is constructed.Then all users are clustered with hierarchical clustering and K-Means clustering,respectively.It is concluded that it is the most suitable to divide the known population into four categories through a comprehensive analysis of the two clustering methods.For the specific clustering results,see the text.