Title: An application of rough set theory to predict telecom customer churn

Authors: Tu Van Binh; Ngo Giang Thy

Addresses: Graduate School, University of Economics Ho Chi Minh City & CFVG, 59C Nguyen Dinh Chieu Street, District 3, Ho Chi Minh City, 700000, Vietnam ' Graduate School, University of Economics Ho Chi Minh City, 59C Nguyen Dinh Chieu Street, District 3, Ho Chi Minh City, 700000, Vietnam; Faculty of Business Administration Nguyen Tat Thanh University, 300A Nguyen Tat Thanh Street, District 4, Ho Chi Minh City, 700000, Vietnam

Abstract: The current paper applies algorithms of machine learning to predict customer churn. The study employs 211,777 instances in the telecommunication sector with six attributes employed, e.g., data, length of stay, top-up, external communication, handset of phone, and churn. Although the rules generation of Naïve Bayes, J48 (Decision Tree), and Decision Table are used, the algorithm of Decision Table is the best candidate to support churn prediction due to its highest accuracy rate of 88.8%. The finding also confirms the role of the external communication of subscribers through calls and messages (in two ways) by other subscribers from the different telecom operators influencing the subscriber's churn. The finding is a significant contribution to the telecom operators to predict churn. In particular, it comes at a time when government regulations have been adjusted to allow phone users to change networks from different service providers, but keep the same phone number.

Keywords: churn; telecom; rough set theory; decision table; Naïve Bayes; decision tree; handset.

DOI: 10.1504/IJCSM.2024.137800

International Journal of Computing Science and Mathematics, 2024 Vol.19 No.3, pp.274 - 284

Accepted: 05 May 2023
Published online: 05 Apr 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article