Machine Learning-Driven Predictive Analytics for Real-Time Supply Chain Risk Management

Abstract

A resilient and efficient supply chain requires real-time risk management in an increasingly volatile global marketplace. This study examines a supply chain risk management system based on machine learning-driven prediction analytics. This research utilizes supervised and unsupervised learning methods, regression models, clustering techniques, and neural networks to improve decision-making, resource allocation, and operational efficiency. Data from logistics, inventory records, and external sources were collected, processed, and analyzed to develop predictive models capable of anomaly detection, forecasting, and planning dynamic responses. Key performance metrics were used in evaluating the proposed system, including MSE, RMSE, MAE, and R². Machine learning models significantly improve supply chain operations, particularly those that use continuous learning and dynamic thresholds. Predictive analytics can transform traditional supply chain management into an intelligent, proactive, and resilient system that enhances performance, mitigates risks, reduces costs, supports rapid decision-making, strengthens responsiveness to disruptions, and effectively addresses uncertainty in highly dynamic and competitive market environments.