Optimizing Procurement Systems Using Classification-Based Supply Delay Predictions

Abstract

Supply delays in procurement can disrupt production operations and impact supply chain efficiency. This study applies data mining classification techniques to predict supply delays in a state-owned fertilizer company in Indonesia. Using the CRISP-DM methodology, we evaluated Logistic Regression, Support Vector Machine (SVM), Random Forest, and Deep Learning models. The results indicate that Deep Learning achieved the highest AUC, demonstrating superior predictive capability, while Random Forest also performed well. Key factors influencing supply delays include order value and Incoterms. These findings can support the development of an early warning system to enhance procurement efficiency and mitigate supply risks.