An Explainable AI Framework for Supply Chain Risk Prediction and Intelligent Decision Support

Abstract

Supply-chain disruptions can affect delivery performance, operational continuity, customer satisfaction, and organizational costs. Conventional risk assessment may have limitations in identifying complex relationships within large operational datasets. Explainable artificial intelligence (XAI) provides an approach for combining predictive analysis with interpretable information to support managerial decision-making. This study aims to develop an explainable AI framework for supply-chain risk prediction and intelligent decision support by integrating machine-learning models with interpretable analytical methods.

The study uses the DataCo SMART Supply Chain dataset. Delivery risk was classified using the Late_delivery_risk variable. Logistic Regression, Random Forest, and XGBoost models were developed using an 80:20 stratified train-test split. Model performance was assessed using accuracy, precision, recall, F1-score, and ROC-AUC. Explainable AI techniques were incorporated to interpret model predictions and identify factors associated with supply-chain risk. The analysis identified substantial variation in delivery-risk patterns across different operational and shipping characteristics. The evaluated machine-learning models demonstrated the ability to distinguish between different risk categories, with XGBoost providing the strongest overall predictive performance. The explainability analysis further supported interpretation of the factors associated with model-generated risk predictions, providing information that can assist supply-chain decision-making. The findings demonstrate that integrating machine learning with XAI can support transparent supply-chain risk prediction and intelligent decision support. The proposed framework provides a structured approach for connecting predictive analytics with interpretable information and proactive supply-chain risk management.