Data-Driven Joint Optimization of Lease Pricing and Container Inventory for Leasing Companies at the Depot Level

Empty container management poses a significant challenge for container depots within global logistics networks, primarily due to demand uncertainty and complex operational costs. This paper addresses the joint dynamic pricing and inventory control problem for a depot managing empty containers over a finite horizon. We propose a data-driven approach that leverages historical operational data to optimize decisions without assuming known demand functions or noise distributions. Our model incorporates key depot-specific features, including the potential dependence of leasing demand on both price and inventory levels, and detailed multi-modal transportation costs for container repositioning. We develop the Uncertain Demand Container Pricing Inventory (UDCPI) algorithm, which utilizes demand hypothesis sets, regularized estimation, and an empirical dynamic programming scheme with sample average approximation. Numerical experiments demonstrate that UDCPI outperforms heuristic data-driven baselines and approaches the performance of a perfect-information benchmark as the data volume in- creases. Sensitivity analyses quantify how pricing bands, inventory–demand interaction, transportation cost structure, and demand volatility affect depot-level profitability. The paper con- tributes a practical, end-to-end data-driven framework tailored to container lessors, together with actionable managerial insights for operating under uncertainty.