Customer Segmentation for Telecommunication Using Machine Learning

With the commercialization of 5G technology and future networks, telecommunications companies need to effectively segment their customers to create targeted strategies that meet their customers’ needs, increase customer satisfaction, and drive revenue growth. Maintaining a competitive position will require diversifying business models in the data plan area. While previous studies have focused on clustering and prediction to identify customer churn, this paper proposes a new approach by utilizing customer segmentation to create different clusters based on usage, behaviour, and loyalty on the Call Detail Records (CDRs) dataset. The segmentation is performed using three distinct approaches, which can be selected based on the needs of the system. A classification algorithm is then trained on the generated labels for each cluster, and the resulting algorithms are evaluated using metrics such as Precision, Recall, F1 score, and accuracy. This novel approach aims to improve customer churn prediction and enable more targeted strategies to retain valuable customers. KNN has shown 96% accuracy for usage and behaviour detection, while ANN managed to get 79% accuracy for the Loyalty of customers.