Data-driven approach for rational allocation of inventory in a FMCG supply chain

The aim of the article is to discuss the major issues concerning forecasting of sales and inventory distribution in traditional grocery retail stores, with a focus on a large supermarket company in Ecuador that operates with more than 200000 SKUs. It aims at the deployment of machine learning algorithms for efficient inventory management so that the business does not experience high stock or low-stock situations. The proposed approach includes the assessment of several supervised machine learning techniques such as Decision Tree, Random Forest, Linear Regression, and XGBoost techniques based on different performance measures that will help to select the best selling forecasting model. These findings underscore the fact that, with high demand uncertainty, heightened market demand rates and supply risks, shifting customer preferences, and ever-reducing product lifecycles, accurate demand forecasting can significantly lower supply chain costs. The study also establishes a need to maintain optimal inventory stock and the distribution of inventory across a number of warehouses. The research implication of the presented study indicates that the machine learning approach advocated for in the research would offer numerous benefits in the management of supply chain for retailers and enhance competitive advantage in the retail industry. To the best of the author's knowledge, this study is novel in its use of sophisticated machine learning approaches to solve problems specific to the grocery retail industry context while also offering a real-world solution to the issues covered.