Document Type : Research Paper

Authors

• Mehdi Seifbarghy ¹
• Shamim Sheikh Ghanbari ²

¹ Department of Industrial Engineering, Faculty of Engineering, Alzahra University, Tehran, Iran

² Department of Industrial engineering, Faculty of Engineering, Alzahra University, Tehran, Iran.

Abstract

Supplier evaluation and optimal order allocation are central challenges in supply chain management. This research addresses these issues through the development of a comprehensive hybrid model that integrates supplier assessment, demand forecasting, and order allocation. Initially, the simple weighted sum method is applied for supplier evaluation, with a novel generalization by using it for both weighting criteria and ranking suppliers. Subsequently, various machine learning algorithms are employed to accurately forecast future demand for supplied items. Finally, a multi-objective optimization model is developed to minimize total supply costs, maximize order allocation to efficient suppliers, and minimize the number of suppliers, subject to probabilistic constraints on the authorized delivery delays and acceptable quality defect levels. These probabilistic constraints are converted into deterministic form using the chance constraint method, and the resulting three-objective model is solved via fuzzy programming through the assignment of membership degrees. Based on the findings, random forest method has the best performance in demand forecasting and the proposed hybrid model can be effectively implemented as an integrated enterprise decision support system.

Keywords

• Supplier selection
• Order allocation
• Simple weighted sum
• Machine learning
• Demand forecasting

Main Subjects

• supply chain management