Data Mining Techniques in Telecommunication Company

Abstract

Due to emerging of amalgam amount of data from variety sources, the data mining has become a hot trend in field of Computer Science. Data mining extracts useful pattern and information from huge amount of existing data with the help of machine learning algorithms that can be helpful in solving many sophisticated problems.

Telecommunication companies also generates big amount of data from providing services to their customers, besides that telecommunication companies suffers from many problems like fraud, Customer churn and …etc.

The generated amount of data from these companies can help them to address the solution for their problems such as Customer Churn. Customer churn indicates to the event when a customer stops using the service of a company and starts to use the service of another company.

Churning of a Customer plays a vital role in having a sustainable business development for a telecommunication company since attracting new customers do not profit a company without retaining the old ones.

Data mining can address the problem by predicting the occurrence of customer churn in Telecom Company, which helps the company to be proactive in this event and can have the chance to retain them before the churn occurs.

In this study, I have chosen two open Telecom Churn data sets and have applied Support Vector Machine, Logistic Regression and Decision Tree Machine Learning Algorithms on each data sets independently, which conclude my work to six experiments.

I have used k-fold cross validation as validation technique during my experiments and confusion matrix for calculating the accuracy of each algorithm, the result of experiments will provide the accuracy of each algorithm in churn prediction for each data set.

At the end we will have a general comparison table from all six experiments which will show the algorithms performance summary and will indicate which algorithm will outperform the others.