Intensified Customer Churn Prediction: Connectivity with Weighted Multi-Layer Perceptron and Enhanced Multipath Back Propagation

In the telecommunication industry, forecasting customer turnover in telecom networks is crucial for maintaining customers and increasing profits. The proposed model explores the importance of predicting churn, emphasizing its financial impact and strategic significance for telecommunications companies. Manual methods for predicting churn are restricted due to their dependence on basic models and human opinion, leading to less than ideal precision and effectiveness. Artificial Intelligence (AI)-based methods have arisen as promising solutions in light of these limitations. Nevertheless, existing AI methods frequently encounter issues like over-fitting, restricted interpretability, and suboptimal generalization. The presented system introduces a novel technique, named "MBP-WMLP” abbreviated as Multipath Back Propagation with Weighted MLP, to improve prediction accuracy in telecommunication networks. The respective approach utilizes a specialized multi-layer perceptron (MLP) structure with weights, designed for managing the intricacies of telecommunications data. Moreover, an improved multipath back-propagation method is presented to improve the optimization of model training and convergence. The projected model combines weighted MLP and enhanced multipath back-propagation to achieve better prediction accuracy, increased interpretability, and improved generalization capabilities. Furthermore, the proposed technique will be applied to authentic Telco customer churn datasets containing sanitized customer activity features and churn labels indicating subscription cancellations for creating predictive models. The projected approach is intended to shows potential for improving customer loyalty tactics and boosting profits for telecom companies in a more competitive market environment.