Home // International Journal On Advances in Telecommunications, volume 10, numbers 1 and 2, 2017 // View article

Modelling and Characterization of Customer Behavior in Cellular Networks

Authors:
Thomas Couronn ́e
Valery Kirzner
Katerina Korenblat
Elena Ravve
Zeev Volkovich

Keywords: Customer behavior pattern; Market segmentation; Probability distribution; Mixture distribution model; Machine learning; Unsupervised classification; Clustering.

Abstract:
In this paper, we extend a model of the fundamental user profiles, developed in our previous works. We explore customer behavior in cellular networks. The study is based on investigation of activities of millions of customers of Orange, France. We propose a way of decomposition of the observed distributions according to certain external criteria. We analyze distribution of customers, having the same number of calls during a fixed period. A segmentation of the population is provided by an approximation of the considered distribution by means of a mixture of several more "basic" distributions presenting the "granularity" of the user's activity. In order to examine the meaning of the found approximation, a clustering of the customers is provided using their daily activity, and a new clustering procedure is constructed. The optimal number of clusters turned out to be three. The approximation is reduced in the optimal partition to a single-exponential one in one of the clusters and to two double-exponential in others. This fact confirms that the proposed partition corresponds to reliable consequential social groups.

Pages: 38 to 49

Copyright: Copyright (c) to authors, 2017. Used with permission.

Publication date: June 30, 2017

Published in: journal

ISSN: 1942-2601