EXPLORING MODEL SELECTION APPROACHES FOR CUSTOMER CHURN PREDICTION IN WEB APPLICATIONS

Customer retention is vital for business sustainability as acquiring new customers is more expensive than retaining existing ones. Predicting customer churn and estimating customer lifetime value (LTV) allows companies to develop targeted retention strategies.This study proposes a model-based approach for churn prediction and LTV estimation in a web application. The literature review examines statistical methods and machine learning algorithms. Several models, including logistic regression, support vector machine, decision tree and random forest, were evaluated for churn prediction, while LTV estimation used linear regression and random forest regressor.The results show that the random forest model achieved the highest accuracy (95.9%) for churn prediction, while the random forest regressor outperformed linear regression in LTV estimation. A segmentation framework combining churn risk and LTV categorizes customers to refine retention strategies. Key predictors such as tenure, cashback amount and delivery distance influenced churn behavior.The proposed approach balances predictive performance and interpretability. Challenges like model generalization remain, suggesting opportunities for future research on broader datasets, transaction data integration and model transparency.