The Role of Artificial Intelligence in Supplier Risk Prediction and Supply Chain Resilience

Supplier delivery problems are a common type of supply chain disruption for which AI applications have been developed to forecast supplier delivery problems․ Prior literature on supplier delivery forecasting has found large differences in forecast accuracy results but little prior research on the value of forecast results to enhance supply chain resilience․ This paper develops a new framework with embedded prescreening and temporal validation steps to forecast and screen supplier delivery risks before shipment and develop resilience scores with limited resources․ An open-access DataCo Smart Supply Chain dataset was used․ The final sample consisted of 172‚765 order observations for 62‚897 orders․ Five models (logistic regression‚ random forest‚ multilayer perceptron‚ XGBoost‚ and LightGBM) were tested and compared with a shipping mode rule using an expanding-window cross-validation design on a later test period of 21‚783 order observations․ XGBoost had an AUC of 0․782 (95% CI 0․773-0․792) and outperformed logistic regression (AUC 0․755) and the rule-based model (AUC 0․748)․ Performance of the XGBoost model was stable over five rolling origins‚ with AUCs between 0․769 and 0․782․ Isotonic calibration of the XGBoost model reduced expected calibration error from 0․022 to 0․013․ SHAP analysis showed that shipping mode‚ scheduled shipping days and order hour were the most important predictors of shipping delays․ Predictive accuracy for standard and second-class shipping methods was no better than chance․ Selecting the 20% of orders with the highest probability of delay identified 34․3% of actual shipping delays correctly‚ with a precision of 0․983․ Findings suggest that AI models can provide useful‚ interpretable predictions of risk to address supply chain problems․ Opportunities for model improvement may exist through additional information on supplier practices.

Authors

Publication Details

Journal
Mathematics for Applications
Published
2026-11-18
DOI
https://doi.org/10.66838/ma.243
Primary Topic
Supply Chain Resilience and Risk Management
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

The Role of Artificial Intelligence in Supplier Risk Prediction and Supply Chain Resilience

Kesava Abisek Rengaraju Senthamil Selvi
3 links
Mathematics for Applications
Supply Chain Resilience and Risk Management
article

The Role of Artificial Intelligence in Supplier Risk Prediction and Supply Chain Resilience

Kesava Abisek Rengaraju Senthamil Selvi
article en

Abstract

Supplier delivery problems are a common type of supply chain disruption for which AI applications have been developed to forecast supplier delivery problems․ Prior literature on supplier delivery forecasting has found large differences in forecast accuracy results but little prior research on the value of forecast results to enhance supply chain resilience․ This paper develops a new framework with embedded prescreening and temporal validation steps to forecast and screen supplier delivery risks before shipment and develop resilience scores with limited resources․ An open-access DataCo Smart Supply Chain dataset was used․ The final sample consisted of 172‚765 order observations for 62‚897 orders․ Five models (logistic regression‚ random forest‚ multilayer perceptron‚ XGBoost‚ and LightGBM) were tested and compared with a shipping mode rule using an expanding-window cross-validation design on a later test period of 21‚783 order observations․ XGBoost had an AUC of 0․782 (95% CI 0․773-0․792) and outperformed logistic regression (AUC 0․755) and the rule-based model (AUC 0․748)․ Performance of the XGBoost model was stable over five rolling origins‚ with AUCs between 0․769 and 0․782․ Isotonic calibration of the XGBoost model reduced expected calibration error from 0․022 to 0․013․ SHAP analysis showed that shipping mode‚ scheduled shipping days and order hour were the most important predictors of shipping delays․ Predictive accuracy for standard and second-class shipping methods was no better than chance․ Selecting the 20% of orders with the highest probability of delay identified 34․3% of actual shipping delays correctly‚ with a precision of 0․983․ Findings suggest that AI models can provide useful‚ interpretable predictions of risk to address supply chain problems․ Opportunities for model improvement may exist through additional information on supplier practices.

Mathematics for ApplicationsVol. 15(2s)
Industry, innovation and infrastructure
Supply Chain Resilience and Risk Management
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon