KPI-Based Analysis and Prediction of Supplier Performance Using Machine Learning

The effectiveness and success of supply chain activities are significantly influenced by the performance of suppliers. In today’s competitive business environment, organizations strive to collaborate with suppliers who consistently meet quality standards, adhere to deadlines, and comply with regulatory requirements. This research investigates the analysis and prediction of supplier performance using key metrics such as quality scores, lead time, compliance, and order status ratios. The study involves the creation of a new quality rating attribute, derived from product, service, and delivery ratings, to assess suppliers’ overall performance over a five-year period (2019- 2023). Through exploratory data analysis, Key Performance Indicators (KPIs) were identified and used to evaluate supplier efficiency and reliability. Machine learning models, including Xtreme Gradient Boost (XGBoost), Naive Bayes, Logistic Regression and Support Vector Machine (SVM) were employed to predict supplier performance based on these KPIs, achieving significant accuracy rates. The results demonstrate that advanced predictive models, particularly SVM, can accurately classify suppliers’ performance, offering valuable insights for businesses to optimize supply chain decisions. In the study, XGBoost achieves 90% accuracy for predicting quality ratings and order status ratios and SVM model achieves 96.66% accuracy in classifying suppliers based on their order status ratio. The study concludes by suggesting future work to incorporate additional supplier attributes and explore advanced machine learning techniques for further improvement in predictive accuracy.