Explainable artificial intelligence-based intuitionistic fuzzy network TOPSIS preference relation models for sukuk investment decisions

• Main Author: Wan Wahida, Wan Mustafa
• Format: Thesis
• Language: English; English; English
• Published: 2025
• Subjects: HG Finance
• Online Access: https://etd.uum.edu.my/11919/1/depositpermission.pdf; https://etd.uum.edu.my/11919/2/s829407_01.pdf; https://etd.uum.edu.my/11919/3/s829407_02.pdf; https://etd.uum.edu.my/11919/

Sukuk investments are increasingly important as Shariah-compliant financial instruments requiring consideration of diverse and conflicting criteria. These challenges require decision support tools that not only handle uncertainty but provide transparent justifications for investment choices. However, despite advancements in artificial intelligence (AI) applications in decision support, explainability in fuzzy multi-criteria decision-making (MCDM) methods such as the technique for order performance by similarity to ideal solution (TOPSIS) remains limited. The lack of explainability reduces trust and transparency, and weakens investor confidence. To address this gap, this study investigates how explainable fuzzy-based MCDM models, can improve decision clarity, reliability, and assurance in sukuk investment decisions. Therefore, the study proposes the development of fuzzy-based explainable models that combine transparency with post-hoc techniques and are aligned with eXplainable artificial intelligence (XAI) properties. Three explainable models were developed, namely, intuitionistic fuzzy network TOPSIS, intuitionistic fuzzypreference relation, and intuitionistic fuzzy network preference relation with a weighting algorithm. The properties of transparency were achieved through model design, while post-hoc explainability was addressed using visualisation, and the models were applied to sukuk data in selecting suitable investment alternatives. The results revealed the ranking of each alternative by preference value and identified the top 15 out of 27 alternatives with corresponding investment proportions. Among these, the highest investment proportions were generated for Alternative 27 (7.08%), Alternative 15 (6.97%), and Alternative 5 (6.88%). The proposed models showed considerable consistency compared to the established model through sensitivity analysis. The visualisation outputs were then validated by decision makers for effectively interpreting the most and least preferred sukuk alternatives. This research contributes to the incorporation of XAI properties into fuzzy MCDM models, introducing a novel explainable model that offers transparent, diversified decisions, validated through decision makers’ feedback. Moreover, this model can also be applied in other domains, extending its relevance beyond sukuk investments