Application Of Machine Learning And Metaheuristic On Demand And Supply Planning Subject To Risks

Abstract

One of the most important steps that determines the overall performance of the supply chain is Planning, which includes demand planning and supply planning. In particular, the new century's big data, which is derived from the intricate customer behaviors, and the fierce competition between players in the same industries have encouraged many companies to invest in the process of supply chain planning. The potential growth of optimization and machine learning algorithms has also enabled experts and researchers to apply these methods to sales forecasting and to create the necessary inventory plans to meet market demand. Three primary objectives are pursued in this essay. This thesis first proposes a sales prediction model using the Extreme Gradient Boosting technique, focusing on the perishability and profitability of products in feature. Secondly, an optimal inventory ordering policy shall be designed with the consideration in discount quantity policies and remaining shelf-life requirement from customers. To provide a complete picture of supply chain risk management for perishable goods, some solutions from the perspectives of business acumen will be proposed in addition to engineering solutions. In this study, the suggested solution is used for the case of Unilever Company, a significant player in the FMCG sector, to help them boost the sales of their powder.