INTEGRATION OF THE SEA-HORSE OPTIMIZER WITH THE ANALYTIC HIERARCHY PROCESS: AN ASSESSMENT OF CONSISTENCY AND ACCURACY

This study integrates the Sea-horse Optimizer (SHO) with the Analytic Hierarchy Process (AHP) to examine whether metaheuristic post-processing of the priority vector improves the consistency and accuracy of AHP weights. The proposed SHO-AHP is compared with two metaheuristic baselines, the Ant Colony Optimizer-Analytic Hierarchy Process (ACO-AHP) and the Genetic Algorithms-Analytic Hierarchy Process (GA-AHP), and with the traditional AHP solution, on a computer-generated 10-factor scenario and on two classic empirical matrices with physically known ground truths, namely Saaty’s geometric area and light intensity examples. Every configuration is replicated across 30 independent runs under an identical budget of objective-function evaluations, and the results are compared using the Friedman test with pairwise Wilcoxon signed-rank tests with Holm correction. Two important findings were discovered. First, the consistency ratio (CR) evaluated at an arbitrary priority vector is not minimized at the principal eigenvector. Expanding the mean-ratio estimate over reciprocal pairs shows that the objective is convex in the logarithms of the weights, with a unique minimizer that lies strictly below the eigenvector value whenever the matrix is inconsistent. All three metaheuristics recover that minimizer and therefore do reduce the reported CR, although the margin is small (0.0204 against 0.0209 in the 10-factor scenario and 0.0807 against 0.0833 in the area example) and a deterministic convex solver reaches the same point in a few milliseconds. Over thirty runs, the ACO-AHP recovers this minimizer most reliably (0.0210 ± 0.0012), whereas the SHO-AHP shows by far the largest dispersion (0.0475 ± 0.0371). Second, consistency and accuracy do not move together. In the area example, the consistency-minimizing vector is 24% closer to the physically known areas than the eigenvector. In contrast, in the synthetic scenario it is 15% farther from the latent priorities that generated the matrix. The practical implication is that lowering this consistency measure is not by itself evidence of a better priority vector. As no human experts were involved in constructing or evaluating the matrices, these findings constitute computational evidence, and validation with expert-derived judgments remains an important direction for future work.

Authors

Institutions

Publication Details

Journal
International Journal of the Analytic Hierarchy Process
Published
2026-11-20
DOI
https://doi.org/10.13033/ijahp.v18i3.1359
Primary Topic
Multi-Criteria Decision Making
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

INTEGRATION OF THE SEA-HORSE OPTIMIZER WITH THE ANALYTIC HIERARCHY PROCESS: AN ASSESSMENT OF CONSISTENCY AND ACCURACY

Ton Nguyen Trong Hien
International Journal of the Analytic Hierarchy Process
Multi-Criteria Decision Making
article

INTEGRATION OF THE SEA-HORSE OPTIMIZER WITH THE ANALYTIC HIERARCHY PROCESS: AN ASSESSMENT OF CONSISTENCY AND ACCURACY

Ton Nguyen Trong Hien
article en

Abstract

This study integrates the Sea-horse Optimizer (SHO) with the Analytic Hierarchy Process (AHP) to examine whether metaheuristic post-processing of the priority vector improves the consistency and accuracy of AHP weights. The proposed SHO-AHP is compared with two metaheuristic baselines, the Ant Colony Optimizer-Analytic Hierarchy Process (ACO-AHP) and the Genetic Algorithms-Analytic Hierarchy Process (GA-AHP), and with the traditional AHP solution, on a computer-generated 10-factor scenario and on two classic empirical matrices with physically known ground truths, namely Saaty’s geometric area and light intensity examples. Every configuration is replicated across 30 independent runs under an identical budget of objective-function evaluations, and the results are compared using the Friedman test with pairwise Wilcoxon signed-rank tests with Holm correction. Two important findings were discovered. First, the consistency ratio (CR) evaluated at an arbitrary priority vector is not minimized at the principal eigenvector. Expanding the mean-ratio estimate over reciprocal pairs shows that the objective is convex in the logarithms of the weights, with a unique minimizer that lies strictly below the eigenvector value whenever the matrix is inconsistent. All three metaheuristics recover that minimizer and therefore do reduce the reported CR, although the margin is small (0.0204 against 0.0209 in the 10-factor scenario and 0.0807 against 0.0833 in the area example) and a deterministic convex solver reaches the same point in a few milliseconds. Over thirty runs, the ACO-AHP recovers this minimizer most reliably (0.0210 ± 0.0012), whereas the SHO-AHP shows by far the largest dispersion (0.0475 ± 0.0371). Second, consistency and accuracy do not move together. In the area example, the consistency-minimizing vector is 24% closer to the physically known areas than the eigenvector. In contrast, in the synthetic scenario it is 15% farther from the latent priorities that generated the matrix. The practical implication is that lowering this consistency measure is not by itself evidence of a better priority vector. As no human experts were involved in constructing or evaluating the matrices, these findings constitute computational evidence, and validation with expert-derived judgments remains an important direction for future work.

International Journal of the Analytic Hierarchy ProcessVol. 18(3)
Van Lang University (VN)
Multi-Criteria Decision Making
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon