Customer Acquisition by Hybrid Intelligence: Explainable HM2CA Framework for Optimizing E-Commerce Marketing with LIME and SHAP

This research proposes a Hybrid Modeller for Maximizing Customer Acquisition (HM2CA), a machine learning framework designed to enhance e-commerce marketing by identifying high-potential buyers. Using the “Online Purchase Intention” dataset of 12,330 sessions, the model integrates SMOTE with Gradient Boosting, achieving 93.47% prediction accuracy, a 3.1 % improvement over the baseline. K-Median clustering segments customers into three categories: High Intention with Revenue, High Intention without Revenue, and Low Intention without Revenue, enabling precise targeting and reducing cart abandonment. Tools like LIME and SHAP enhance interpretability, explaining feature importance and classification reasoning. This transparency allows marketers to refine strategies effectively. By combining classification, clustering, and interpretability, the HM2CA framework optimizes marketing efforts, improving ROI and efficiently identifying high-value customers.