AI-Driven Supply Chain Analytics: Leveraging Machine Learning and Predictive Analytics for Risk Identification and Decision Support

Abstract

As supply chain disruptions continue to escalate, they underscore the growing fragility of global supply chain networks as a result of geopolitical conflicts, demand volatility, transportation challenges, and environmental volatility. The problems are a result of a narrow definition of risk that is based on manual analysis and historical data.The problems are manifestations of a narrow definition of risk, which is based on manual analysis and historical data. In complex supply chain environments, machine learning is a promising methodology that can be used to build more intelligent predictive systems and help make better decisions. This research aims at developing and testing a machine learning-based risk prediction model that can detect possible disruptions in the supply chain and allow to detect critical risk factors in logistics operations on the global level in time. The data from the supply chain were combined with disruption indicators from international logistics activities and a quantitative experimental approach was used. The data set comprised 5,420 active records from 2018 to 2024 from publicly available global logistics databases and enterprise risk management servers. Several machine learning algorithms were used and compared:Random Forest, Gradient Boosting and Support Vector Machines. Experimental results showed that Gradient Boosting algorithm had the best predictive performance with 94.2% accuracy. The model was able to identify significant factors for supply chain risk, such as demand variability, supplier reliability, and transportation delays, through the results of the multiple parameters. The results validate the use of machine learning predictive models to boost supply chain resilience by detecting risks in advance and likely supporting decision-making processes..