Data-driven reliable decision-making approach for last mile delivery to handle uncertainty: challenges and opportunities

The last-mile delivery phase, a critical component of the supply chain, faces increasing complexities driven by the rapid growth of e-commerce and urbanization. This chapter explores a data-driven approach to addressing the uncertainties inherent in last-mile logistics, including fluctuating customer demand, traffic congestion, and environmental factors. Traditional vehicle routing methods are no longer sufficient to handle the dynamic nature of modern delivery operations. Instead, advanced optimization techniques, such as stochastic dynamic vehicle routing problems (SDVRP) and Markov decision processes (MDP), offer robust solutions. By leveraging real-time data, machine learning, and predictive analytics, these methodologies enable logistics providers to make adaptive decisions, improving operational efficiency and customer satisfaction. This chapter demonstrates how a data-driven approach can effectively manage the unpredictability of last-mile delivery, contributing to more flexible, sustainable, and responsive logistics systems in the increasingly complex landscape of urban environments. © 2026 Elsevier Inc. All rights reserved.