Detection of hidden attacks to telecommunication company on the basis of the data mining technologies

Abstract

In this article we consider an approach of analysis the telecommunications company customers information using data mining technology. The purpose of this work is to develop a software solution for identifying customers who behave abnormally compared to ordinary customers. To solve the problem the method of cluster analysis is used, with k-means, method of support vector machines (SVM) and nonstandard method of initializing k-means++ centroids. With the application of the developed programs, the analysis of the data obtained and the identification of customers who are suspected of hidden attacks on the company are identified. To classify the new data the classifier is developed, which is trained on large data sets and based on SVM.