Abstract

Customer churn is a significant challenge in the telecommunications industry, particularly in competitive markets such as the United Arab Emirates, where customers can easily switch between providers such as e& and du. Telecommunications companies must move beyond reactive approaches and adopt data-driven strategies that enable early identification of churn risk and support proactive retention. This study develops an interpretable churn prediction framework using logistic regression and decision tree models within IBM SPSS. Following the CRISP-DM methodology, the research identifies key demographic, behavioural, and service-related factors that influence churn, and evaluates model performance using accuracy, precision, recall, and ROC-AUC. The findings confirm that churn is driven primarily by contract type, internet service type, payment method, and value-added service adoption rather than demographic characteristics. The full-profile logistic regression model achieved an accuracy of 94.9% and an AUC of 0.977, demonstrating that interpretable models can deliver strong predictive performance while remaining accessible to business teams. The model outputs translate directly into actionable retention strategies, including targeting month-to-month contract holders, promoting value-added services, and incentivising automatic payment adoption. This study contributes a transparent and practically deployable churn prediction framework tailored to telecommunications environments, with future work recommended to validate the approach using real operational data from UAE providers.

Publication Date

5-2026

Document Type

Thesis

Student Type

Graduate

Degree Name

Professional Studies (MS)

Department, Program, or Center

Graduate Programs & Research

Advisor

Hammou Messatfa

Campus

RIT Dubai

Download

Share

COinS