Enhancing Predictive Accuracy for Customer Churn in Digital Banking: A Multi-Model Analysis

Customer churn prediction has become a crucial aspect for digital banking institutions that are focusing mainly on enhancing customer retention and improving their marketing strategies. This research tries to compare the effectiveness of three machine learning models i.e. Logistic Regression, XGBoost, and Neural Network in analyzing its prediction for customer churn using the dataset of 27,000 entries with 31 different questions containing demographic and transactional data. The results obtained disclosed that out of the three ML technique XGBoost model was performing better than the other two models with an accuracy level of 73.40% with weighted F1-score of 0.73. Cross-validation with the CV Score (0.7314) and Low SD further confirmed the quality of XGBoost model as consistent performer. The findings provide valuable insights for financial institutions to develop more effective customer retention strategies for a better customer approach.