A BIG DATA-DRIVEN ANALYTICAL FRAMEWORK FOR PREDICTING CUSTOMER PRODUCT PREFERENCES AND CHURN IN NIGERIA

Abstract

The rapid growth of Big Data has transformed organizational decision-making and operational efficiency across industries, particularly in the telecommunications sector, where understanding customer behaviour is crucial for reducing churn and optimizing revenue. This study presents a Big Data-driven analytical framework for predicting customer product preferences and churn in MTN Nigeria, leveraging a dataset of 974 anonymized customers with 158,943 transactional records collected between 2024 and early 2025. The methodology integrates data preprocessing, normalization, class imbalance correction using SMOTE, and supervised machine learning algorithms, including Random Forest (RF), Backpropagation Neural Network (BPNN), K-Nearest Neighbours (KNN), and Naïve Bayes (NB). Models were trained and evaluated using accuracy, precision, recall, and F1-score metrics, with RF achieving the highest overall performance (accuracy = 89%, F1-score = 0.94). The framework was implemented as a Python-based prototype system with modules for customer segmentation, product targeting, churn risk prediction, and personalized marketing communication. Results demonstrate that the proposed system can accurately predict customer preferences, reduce unsolicited promotions, and improve customer retention and revenue. This study highlights the potential of integrating Big Data analytics and machine learning into telecom CRM workflows and provides a scalable model applicable to other sectors requiring behaviour-driven customer engagement strategies.