Abstract
The demographic and geographic variables that influence customer attrition in the telecommunications sector are investigated in this study. To find complex connections between sociodemographic variables, geographical traits, and consumer behaviour, a multilayer perceptron (MLP) model was utilised. Age and income are important variables associated with customer attrition, according to an analysis of 51048 customers. Additionally, the results show that rural clients are more likely to experience churn. In terms of accuracy, recall, and other performance criteria, the multilayer perceptron fared better than conventional machine learning techniques. The study suggests focused retention tactics suited to particular demographic and geographic client segments in light of these findings. To support efforts to increase revenue and profitability, it is advised that deep learning techniques, like the multilayer perceptron, be used in real-world business applications to uncover variables driving customer turnover.