An Improvement Method to Mining Outliers Based on Social and Spatial Information

Abstract

In this paper, a method for finding outliers is proposed for communication data. By extracting the social and spatial information as the eigenvalues from the user’s call list. We use clustering method to aggregate users with the same attributes, and then analyze the clusters to find outliers. We use the real communication data of a communication company as the data set, and the experimental results show that our method can quickly and accurately find the abnormal communication user.