E-commerce Logistics and Delivery Risk Management with Machine Learning and MCDM Methods

Abstract

Machine learning techniques and fuzzy Multi-Criteria Decision Making (MCDM) are sophisticated methodology widely employed to support data analysis and decision-making in systems characterized by uncertainty and vagueness. Given the inherent imprecision present in real-world data, these approaches leverage fuzzy logic to effectively address ambiguities and provide meaningful insights. In this thesis, we incorporate the machine learning technique with fuzzy cubic MCDM methods which helps to reduce the complexity in large scale MCDM problem. In particular the fuzzy C--Means algorithm is employed to reduce large data into meaningful clusters. Then these clusters are ranked and evaluated using MCDM techniques, specifically enhanced by fuzzy cubic numbers. The proposed algorithm allows decision-makers to effectively manage uncertainties inherent in decision-making processes that further improve the evaluation process including EDAS (Evaluation Based on Distance from Average Solution) and MAIRCA (Multi-Attributive Ideal-Real Comparative Analysis). Through the incorporation of fuzzy cubic numbers, these methods offer a precise evaluation that corresponds with the uncertainties and complexities of the real world. We also apply this proposed algorithm for risk assessment and mitigation in e-commerce logistics to find the best delivery routes for deliver different products. This hybrid strategy not only helps to identify delivery routes that best combine operational effectiveness and risk management but will also increase decision-making accuracy.