Integration of association rule mining and RFM analysis with machine learning for e-commerce customer value segmentation: a sustainable retail perspective
摘要
This study presents a novel framework combining association rule mining, RFM analysis, and machine learning to improve customer value segmentation in e-commerce. Using digital transaction data, the approach evaluates behavioral patterns and sustainability engagement. Validated through a sustainable food e-commerce retailer, the framework distinguishes high-lift from non-high-lift customer segments. An ensemble model, integrating Random Forest and XGBoost, achieves 97.19% accuracy and an AUC score of 0.9953. Feature importance analysis highlights behavioral metrics, particularly purchase frequency, as key predictors of customer value, challenging traditional demographic-based methods. Notably, a new link between digital sustainability engagement and customer value is revealed, with high-lift customers excelling in ESG performance. This research enhances understanding of customer value in digital retail and offers actionable strategies for precise segmentation and targeted marketing, supporting both customer value creation and sustainability in the evolving e-commerce landscape.