Enhancing Customer Retention through Feature Selection and XGBoost Classification for Churn Prediction

Customer churn remains a significant and pressing challenge for both large enterprises and businesses across various industries. Particularly within the banking sector, where revenue is directly influenced by customer retention, there is a concerted effort to enhance predictive methods for identifying potential churn. This paper explores different machine learning (ML) algorithms employed in constructing a churn model to assist banking operators in predicting customers at risk of churning using minimum attributes and aiming to achieve good accuracy. The research delves into experimental results to compare the efficacy of various techniques, identifying the model that works best for predicting customer attrition in the banking sector in the end. This study uses feature selection and the XGBoost algorithm to select attributes based on their ranks.