Machine Learning-Based Supplier Risk Prediction and Resilience Scoring for Strategic Sourcing in U.S. Manufacturing Supply Chains

The increasing complexity and global interdependence of U.S. manufacturing supply chains have significantly amplified supplier related risks, including disruptions arising from financial instability, operational failures, geopolitical tensions, and natural disasters. Traditional supplier evaluation approaches largely rely on static, rule-based scorecards and expert judgment, which are insufficient for capturing dynamic risk evolution and resilience behavior in modern supply networks. This paper proposes a machine learning based framework for supplier risk prediction and resilience scoring to support strategic sourcing decisions in U.S. manufacturing supply chains. Using a multi source dataset comprising financial indicators, operational performance metrics, logistics reliability measures, and external disruption signals, the proposed framework applies supervised learning models to predict supplier risk levels and constructs a composite resilience score through normalized and entropy weighted indicators. The methodology integrates data preprocessing, feature engineering, model training, validation, and hypothesis based statistical testing. Experimental results demonstrate that the proposed approach significantly outperforms traditional scorecard-based risk assessment methods in predictive accuracy, robustness, and early warning capability. Statistical significance tests confirm the reliability of the observed improvements. The findings highlight the potential of data driven supplier intelligence systems to enhance supply chain resilience and strategic sourcing effectiveness. Limitations, future research directions, and managerial implications are discussed.