A Novel Three-Way Decision Framework for Classifying Spare Parts between Additive and Conventional Manufacturing

This paper presents an enhanced Three-Way Decision (TWD) framework designed to classify spare parts production into Additive Manufacturing (AM), Conventional Manufacturing (CM), or uncertainty, thereby improving decision-making accuracy and adaptability. The approach integrates logistic regression-based conditional probability estimation with relative loss functions that evaluate the costs associated with AM, CM, and inventory-related risks. First, logistic regression is applied to dynamically estimate the conditional probabilities of each spare part being assigned to AM or CM. Second, a relative loss matrix is calculated, which is based on the costs associated with AM, CM, and inventory-related risks. Finally, the framework calculates the expected loss for each spare part and categorizes them into three decision regions: AM, CM, or uncertainty. A case study on spare parts production validates the model's effectiveness, demonstrating its ability to enhance risk management and decision-making precision in practical scenarios.