A proposed framework for supplier selection and order allocation using machine learning clustering and optimization techniques

The process of selecting the most suitable suppliers and allocating orders to them is critical in supply chain management. This research proposes a new framework to address the challenges of Supplier Selection and Order Allocation (SS&OA) by introducing a two-phase combined approach. In the first phase, three Machine Learning (ML) clustering techniques (i.e., K-means clustering, Gaussian Mixture Model, and Balance Iterative Reducing and Clustering using Hierarchies) are employed to identify a small group of suitable suppliers from a large pool of potential suppliers. The accuracy of the developed clustering models is assessed using Silhouette score technique. In the second phase, we focus on one of the clusters based on the results of Phase 1. In this phase, a new multi-objective model is developed for SS&OA that considers multi-source, multi-period, and multi-product scenarios. Compromise method is utilized to obtain efficient solutions. The framework is applied to extensive, real historical contract data from Canada's Public Works and Government Services Canada (PWGSC) on behalf of federal departments and agencies. The results indicate that K-means clustering model is the most accurate among the models examined for this dataset, and the choice of ML clustering techniques has a significant impact on SS&OA.