Decision support for integrated trade agent's procurement and sales planning under uncertainty

Abstract

This paper investigates a trade agent decision optimization problem (TADOP), in which a trade agent (TA) selects a subset of retailers and suppliers to maximize its profit under uncertain demand and spot price. The TA operates between suppliers and retailers as a third-party platform and decide which subset of retailers to serve, taking into account capacity reservations with option suppliers in advance. Once demand and spot price are realized, the TA decides how much to procure from each channel to fulfill retailers' demand. The problem is formulated as a two-stage stochastic program. Due to the high complexity and large number of scenarios, we reformulate the problem as a set-partition model, where the master problem (MP) selects the combination of retailers to serve, and the subproblem (SP) identifies the optimal procurement plans, thus reducing the number of variables and constraints. To further enhance tractability, the SP is transformed into an equivalent shortest-path problem (SPP) to address issues of non-linearity and non-convexity. Experimental results demonstrate the effectiveness of the decomposition approach, providing TAs with a practical decision-making tool for procurement and sales. Furthermore, the insights gained into TAs' procurement and sales strategies across various scenarios offer valuable guidance for decision-making in uncertain supply chain environments.