Particle Swarm Optimization Based-Algorithm For An Optimal Completion Of Incomplete Pairwise Comparison Matrices

Abstract

This study introduces a method based on Particle Swarm Optimization (PSO) to estimate the missing entries in incomplete Pairwise Comparison Matrices (PCMs), commonly utilized in multi-criteria decision-making processes. Incompleteness, due to missing data or subjective biases, often affects the consistency and reliability of decisions. The proposed method refines the estimates iteratively to minimize the consistency ratio and improve matrix coherence. Numerical experiments show that PSO offers a robust and efficient solution. A comparison with the Geometric Mean Method (GMM) across various matrix sizes demonstrates that PSO achieves higher accuracy, particularly in terms of consistency and estimation quality.