EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Endogenous Reliability Calculus of Objective-Weighted MCDM: Differential Sensitivity and Certified Rank Robustness

Ibrahim M. Hezam

Informatica · 01.01.2026 · 1-33

DOI: 10.15388/26-infor649 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Informatica, online erschienen am 20.11.2026. Verfasst von Ibrahim M. Hezam. Die Arbeit nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Objective weighting in multi-criteria decision-making (MCDM) makes criterion weights functions of the same performance data used to rank alternatives. Data uncertainty therefore propagates through both normalized performances and weight re-estimation. This paper develops an endogenous reliability calculus for the complete raw-data-to-ranking pipeline. The composite Jacobian is formulated for general differentiable aggregation, while closed-form pairwise and certification results are developed under additive aggregation. Pairwise sensitivity is decomposed exactly into a direct performance component and an endogenous reweighting component, yielding local conditions for amplification, attenuation, neutrality, and cancellation. A Conditioning Transfer Principle further shows that ill-conditioned objective weights need not imply an ill-conditioned decision because weight and ranking sensitivities are different projections of the same information Jacobian. The analysis then moves from local diagnostics to finite guarantees. A deterministic raw-data certificate is derived for rank preservation; information-separable weighting permits exact criterion-wise decomposition, whereas coupled rules with a suitable factorization admit a monotone factor-clustering hierarchy. Standard deviation, Entropy, and CRITIC instantiate local-dispersion, information-divergence, and network-coupled structures, respectively. Numerical experiments verify the analytical Jacobians, distinguish weight conditioning from decision conditioning, and show amplification, near-neutrality, and local attenuation. The worked SD example additionally certifies the unique winner for every componentwise relative perturbation up to 7.210%, while an explicit adverse perturbation produces a reversal at 7.211%. The framework therefore shifts robustness analysis from independent parameter perturbation toward reliability analysis of the complete endogenous decision mechanism.

Zitieren

APA

Hezam, I. M. (2026). Endogenous Reliability Calculus of Objective-Weighted MCDM: Differential Sensitivity and Certified Rank Robustness. Informatica, 1-33. https://doi.org/10.15388/26-infor649

Vancouver

Hezam IM. Endogenous Reliability Calculus of Objective-Weighted MCDM: Differential Sensitivity and Certified Rank Robustness. Informatica. 2026:1-33. doi:10.15388/26-infor649

BibTeX
@article{Hezam2026,
  author = {Hezam, Ibrahim M.},
  title = {Endogenous Reliability Calculus of Objective-Weighted MCDM: Differential Sensitivity and Certified Rank Robustness},
  journal = {Informatica},
  year = {2026},
  pages = {1--33},
  doi = {10.15388/26-infor649}
}
Kostenlos starten →