A CatBoost-Based integrated decision-making model for knowledge-oriented risk evaluation in multi-tier supply chains

This study proposes a data-driven integrated framework based on knowledge management for risk evaluation in multi-tier supply chains. Traditional risk assessment models often rely on qualitative or expert-based methods, limiting their adaptability and accuracy in dynamic environments. The proposed framework combines operational risk indicators (supplier, production, and distribution) with knowledge management indicators (creation, sharing, application, retention, and market updating) to provide a holistic understanding of risk propagation. Using a dataset of 800 records and the CatBoost machine learning algorithm, the model achieved an accuracy of 0.86 and a weighted F1-score of 0.85. Sensitivity analysis showed that delivery delay (SR1), price volatility (SR2), and knowledge application (KM3) are the most influential factors, highlighting the interdependence between operational efficiency and knowledge utilization. The findings demonstrate that integrating knowledge management with data-driven analytics enhances prediction accuracy and strengthens organizational resilience. This research contributes a quantitative and explainable model for proactive risk management and provides managers with actionable insights to build adaptive, knowledge-oriented supply chain systems.