Advanced Optimization in E-Commerce Logistics: Combining Matheuristics With Random Forests for Hub Location Efficiency

Abstract

The rapid growth of e-commerce has introduced significant challenges in optimizing logistics and distribution networks, particularly in determining hub locations to balance cost efficiency and service quality. This study presents a novel framework integrating matheuristics with Random Forest machine learning to address the Hub Location Problem (HLP) in e-commerce logistics. Leveraging real-world data from a global logistics company, the methodology combines predictive analytics with robust optimization techniques to dynamically adjust hub configurations based on fluctuating demand and operational constraints. Experimental results demonstrate substantial improvements over traditional methods, including an improvement of 7.8% in delivery time compliance, a 9.4% increase in fulfillment rate and enhanced network robustness by 16%. The findings highlight the strategic advantages of integrating advanced analytics into logistics planning, offering actionable insights for industry practitioners.