Collaborative Forecasting with Reinforcement Learning to Enhance Resilience

As the worldwide business environment is getting more complex, supply chains need to enhance resilient operations to face new troublesome, disruptive events, such as significant changes in sourcing, inventory management, and production planning. Supply Chain Collaboration has been an important strategic action to achieve resilience; however, it still has many challenges. One major issue is forecasting collaboration, where reaching common goals has proven difficult. Several companies have experienced diverse forces and directions depending on individual collaborators’ goals, leading to weak forecasting processes that hinder the alignment of all the functional plans throughout the supply chain. In this research, we propose a reinforcement learning model to help demand planners select the best collaboration inputs from other stakeholders to improve forecasting accuracy, help maintain a balance between supply and demand and increase resilience in the planning process. The main objective of this article is to exemplify how companies can use artificial intelligence to improve collaborative forecasting and foster enterprise and supply chain resilience. Moreover, an essential contribution of the present research is the enhancement of collaborative forecasting through reinforcement learning and deploying its combined techniques in an accessible web tool for practitioners and academics. Further research will involve the complete implementation of the proposed collaborative tool in, at least, one company already engaged in internal sales and operations planning or collaborative planning and forecasting replenishment.