Paper Title
Inference-Based Demand Forecasting in Supply Chains

Abstract
In this study, we employed mathematical modeling and simulation as quantitative tools to address downstream demand inference and enhance supply chain performance. Initially, we demonstrate the feasibility of demand inference at any downstream level under specific assumptions, providing a more comprehensive visibility into demand propagation along the supply chain. Subsequently, a bi-objective genetic algorithm is proposed to concurrently improve upstream average inventory and mitigate the bullwhip effect. To validate the effectiveness of the approach, we conducted data collection based on simulated ARMA models. Optimization results reveal a significant reduction in the bullwhip effect when specific weightings are selected from the Pareto frontier. Furthermore, we aim to model the intricate relationship between the two studied metrics, a task proven complex in identifying the most suitable model. Additionally, a two-way robustness analysis is performed to examine the reliability of the adopted framework, demonstrating that supply chain parameters and algorithm initialization do not impact the bullwhip effect variation. This study suggests diverse possibilities for collaborative forecasting in supply chains, such as local or global collaboration. Keywords - Supply Chain Management, Downstream Demand Inference, Weighted Moving Average Method, Bullwhip Effect, Multi-Objective Optimization, Genetic Algorithm