Community Detection Approach in Telecommunication Clients Grouping

Abstract

In this thesis, we are tackling a community detection problem (CDP); specifically clustering of customers of some telecommunication company in the UAE. CDP can be defined as an optimization problem by using mathematical programming. The objective function we work with in this thesis is modularity maximization. The method for solving this problem is based on metaheuristic Variable Neighborhood Search (VNS). Real data from the telecommunication company is collected and prepared as an input to be inserted in the VNS code. Two different types of input data graphs were prepared and tested under equal parameter values. The results are evaluated using sensitivity analysis using scatter plots, comparison of results using histograms, and visualization using python plotter to draw map plots.

Date of Award	Jul 2022
Original language	American English