In e-commerce, robust multi-criteria decision analysis is essential for accurately aligning rankings with customer preferences. Addressing this need, the Fuzzy Normalization-based Multi-Attributive Border Approximation Area Comparison (FN-MABAC) method introduces an advanced extension of the traditional MABAC model. The primary objective is to mitigate the ranking reversal paradox while enhancing decision-makers’ preference representation.To achieve these goals, FN-MABAC employs a fuzzy normalization approach, providing a structured framework that minimizes reversal risks and aligns with reference rankings. Experimental analyses compared FN-MABAC with established methods, including Technique of Order Preference Similarity to the Ideal Solution (TOPSIS) and Characteristic Object METhod (COMET), using datasets on graphics card evaluations. Results demonstrate that FN-MABAC achieves higher concordance with reference rankings and exhibits superior resistance to ranking reversals compared to traditional approaches. Specifically, FN-MABAC achieved a Weighted Similarity (WS) coefficient match level of 0.97522 with the reference ranking, while STFN-MABAC reached a level of 0.99599, highlighting the high potential of these methods to adapt to preferences in the presented e-commerce example.Conclusions drawn indicate FN-MABAC’s potential to offer precise, stable rankings essential for e-commerce decision-making. Additionally, the concept of Stochastic Fuzzy Normalization-based MABAC (STFN-MABAC) is introduced as a re-identification approach, further aligning ranking outputs with customer preferences, particularly in scenarios with incomplete data. This study underscores FN-MABAC’s significance as an effective, resilient tool in multi-criteria analysis, contributing to improved decision-making within e-commerce contexts.
Read full abstract