Dynamic Swing Weights Method for Supplier Selection: A Hybrid Approach Integrating Expert Judgement and Machine Learning

Abstract

Selecting the right suppliers is a crucial part of supply chain management. Choosing a supplier is a difficult strategic choice that involves weighing a number of potentially conflicting factors, and it may alter over time as providers' real performance improves. Fixed weights derived mostly from expert judgment are used in many classic multi-criteria decision-making techniques, which might add bias and reduce the outcomes' responsiveness to changes in operational conditions.The Dynamic Swing Weights Method (DSWM), a hybrid approach, is presented in this article. After obtaining initial expert opinions using the Swing Weights method, the decision weights are updated over time using SHAP-based explainable machine learning model. The model allows the weights to fluctuate as conditions change by combining operational data with expert judgment.This case study demonstrates that DSWM performs better than static techniques in terms of classification accuracy and change adaptation speed using simulated industrial data. By serving as a transparent and scalable decision support system, the proposed structure satisfies the needs of modern supply chains.