Supplier Segmentation Method Using Supervised Machine Learning: A Case Study of Qatar Foundation

Abstract

A buying strategy involves efficient supplier segmentation to deal with each set of suppliers adequately, as not all suppliers can be managed with the same approach, especially in the case of a high number of suppliers. Moreover, the majority of the currently available supplier segmentation strategies either rely on subjective judgements or demand a great deal of work. To address this issue, this study proposes a supplier segmentation strategy using the Kraljic model and supervised machine learning (ML) techniques to group suppliers into four segments: leverage, non-critical, strategic, and bottleneck. The methodology involves applying supervised ML techniques to actual purchase data for the Qatar Foundation, as it has huge transactions and suppliers due to its nature and uniqueness. Five different ML models (boosting, decision tree, k-nearest neighbors, random forest, and support vector machine) are used and evaluated for their performance. The best model is found to be the random forest algorithm, which showed high accuracy on most evaluation metrics. Using this automated supplier segmentation method, the procurement team can reduce the time and effort required for supplier management. The proposed approach provides a more objective and data-driven method for supplier segmentation than traditional methods, which often depend on subjective decisions or require significant effort. This study offers valuable insights for organizations looking to improve their supplier management practices and increase efficiency in their procurement processes. Keywords: supplier segmentation, supervised machine learning, supply chain management, supplier relationship management, procurement.

Date of Award	2023
Original language	American English
Awarding Institution	HBKU College of Science and Engineering