MACHINE LEARNING–DRIVEN SMART SHARING PLATFORMS FOR OPTIMIZING LAST-MILE DELIVERY IN SAUDI ARABIA’S E-COMMERCE SECTOR

Background: Last-mile delivery accounts for over 50% of total logistics costs, representing a major operational and sustainability challenge in Saudi Arabia’s rapidly expanding e-commerce sector. Conventional siloed fleet models often result in underutilized capacity, delivery inefficiencies, and elevated carbon emissions. This study proposes a machine learning–driven smart-sharing framework that integrates predictive analytics with operations research to optimize last-mile delivery in the Kingdom of Saudi Arabia. Methods: Using the Regional Delivery Data in Saudi Arabia (2024), which aggregates quarterly order volumes across 13 regions, this study develops demand forecasting models (SARIMAX, LightGBM, and CatBoost) as well as ETA predictions enriched with event-and calendar-based features. These outputs inform a vehicle routing problem with time windows (VRP-TW) implemented using Google OR-Tools and a CP-SAT facility location model for parcel locker siting. A digital twin simulation built in SimPy is then used to stress test performance under peak demand conditions (e.g., Ramadan) and weather-related disruptions. Results: The findings show that pooled fleet sharing reduces routing costs by more than 65%, decreasing traveled distance from 7,906 to 2,370 units, while increasing the share of on-time deliveries from 37% to nearly 100%. The introduction of parcel lockers further improves system efficiency, reducing kilometers traveled per parcel by up to 90% and lowering CO₂ emissions from 0.77 to 0.06 kg per parcel. Stress-test experiments confirm the resilience of the shared model, which maintains service-level agreement (SLA) compliance even under demand surges and operational disruptions. Conclusions: This research presents the first reproducible, Colab-native machine learning and optimization pipeline applied to Saudi Arabia’s regional delivery data. The proposed framework provides actionable insights for logistics managers, urban planners, and policymakers seeking data-driven approaches aligned with Saudi Arabia’s Vision 2030 sustainability and efficiency goals.