Predictive CLV modeling for personalized CRM strategies in retail

This thesis examines the predictive modeling of Customer Lifetime Value (CLV) in the context of non-contractual, high-frequency offline retail to fill out the research gap in the existing empirical research, which often overlooks this specific segment of retail. The thesis aims to address these limitations in the literature by analyzing the possible approaches to the prediction of CLV in this specific segment of retail and identifying the most appropriate method of predicting the customer lifetime value by empirical research conducted on the data of the real company operating in this industry. Through the application of machine learning models, such as XGBoost and Random Forest, the study highlights their capacity to manage complex data and customer variability, making them particularly suitable for retail environments. The research offers a step-by-step framework for CLV modeling, covering data preparation, model training, and evaluation while presenting actionable insights for business managers.