Supplier Selection and Relationship Management: An Application of Machine Learning Techniques

Abstract

Managing supply chains is an extremely challenging task due to globalization, short product life cycle, and recent advancements in information technology. These changes result in the increasing importance of managing the relationship with suppliers. However, the supplier selection literature mainly focuses on selecting suppliers based on previous performance, environmental and social criteria and ignores supplier relationship management. Moreover, although the explosion of data and the capabilities of machine learning techniques in handling dynamic and fast changing environment show promising results in customer relationship management, especially in customer lifetime value, this area has been untouched in the upstream side of supply chains. This research is an attempt to address this gap by proposing a framework to predict supplier future value, by incorporating the contract history data, relationship value, and supply network properties. The proposed model is empirically tested for suppliers of public works and government services Canada. Methodology wise, this thesis demonstrates the application of machine learning techniques for supplier selection and developing effective strategies for managing relationships. Practically, the proposed framework equips supply chain managers with a proactive and forward-looking approach for managing supplier relationship.