Transforming Supply Chain Operations through AI and Machine Learning: Optimizing Demand Forecasting, Inventory Management, and Logistics Efficiency

Abstract

This paper examines the application of Artificial Intelligence (AI) and Machine Learning (ML) technologies in optimizing demand forecasting, inventory management, and logistics within online retail supply chains. Through a comprehensive dataset of a UK-based retailer (541,909 records of transactional data) the paper explores the possibility of AI/ML algorithms to optimize the effectiveness and accuracy of these important supply chain factors. The study presents a new two-stage Customer-to-Inventory (C2I) framework based on DBSCAN to separate customers and tests a variety of forecasting models, such as Long Short-Term Memory (LSTM) networks, Prophet and such classical forecasting models as Linear Regression and Random Forest. The findings indicate that Linear Regression model is more effective than the other two in demand forecasting with RMSE of 23,120 which is still much lower in terms of prediction error than Prophet (31,613) and Random Forest (23,880). Moreover, demand forecasting and inventory management algorithms based on AI result in a 15% optimization in stocks levels, which minimizes stockouts and overstocking. It is also noted in the study that customer segmentation via DBSCAN identifies the high-value customers who are contributing 60% of total sales, which can be used as actionable information to use in targeted marketing and tailored promotions. This study provides a viable blueprint of the deployment of models of e-commerce companies, the findings of which can be used to increase the efficiency of operations, minimize expenses, and increase customer satisfaction.