Extracting offline retail shopping patterns: a restricted Boltzmann machines approach to customer segmentation and cross-selling

초록

This study introduces restricted Boltzmann machines (RBM) as a novel approach for customer segmentation and cross-selling optimization in offline retail environments. By extracting interpretable latent shopping patterns from sparse transaction data, RBMs effectively capture nuanced multi-category purchase behaviors by modeling unobservable customer preferences as latent factors. Unlike traditional clustering-based segmentation methods, our RBM approach reveals more meaningful cross-buying patterns that directly inform targeted cross-selling strategies. Our key contributions include: (1) demonstrating RBMs’ superior ability to identify actionable segments based on cross-buying patterns compared to benchmark clustering methods, and (2) demonstrating that integrating RBMs into collaborative filtering models significantly improves predictive performance for cross-selling recommendations. The results highlight RBMs’ dual effectiveness in creating interpretable customer segments while enabling data-driven cross-selling strategies that address the inherent sparsity challenges of offline retail transaction data. © 2025 Elsevier Ltd

키워드