Enhancing Supply Chain Agility and Sustainability through Machine Learning: Optimization Techniques for Logistics and Inventory Management

Abstract

As global supply chains face increasing complexity, the demand for agile and sustainable management strategies has become more critical. This study employs advanced machine learning (ML) techniques to transform logistics and inventory management, moving beyond the constraints of traditional analytical methods. Utilizing historical data from a multinational retail corporation, including sales, inventory levels, order fulfillment rates, and operational costs, we have applied a range of ML algorithms such as regression, classification, clustering, and time series analysis. These models were developed to tackle key operational challenges, enhancing decision-making by improving demand forecasting accuracy by 15%, optimizing stock levels by reducing overstock and stockouts by 10%, and predicting order fulfillment timelines with 95% accuracy. Additionally, our approach enabled the identification of at-risk shipments and the segmentation of customers based on their delivery preferences, facilitating personalized service offerings. A comprehensive evaluation of these models showed significant improvements in predictive accuracy, efficiency in lead time by 12%, silhouette coefficients for clustering at 0.75, and a reduction in replenishment errors by 8%, highlighting the transformative potential of ML in making supply chain operations more responsive and data driven.