Road safety measurement with reliability using an advanced hybrid decision model

Abstract

Abstract This study proposes a brand-new hybrid multi-criteria decision-making (MCDM) framework that combines High-Dimensional Vector Projection (HDVP) and Between-class Variance Maximization (BeVarMax), termed the HDVP–BeVarMax model, aiming to provide trustworthy decisions and defensible policy conclusions. Specifically, HDVP quantifies the relative proximity of each country to an ideal performance vector in a high-dimensional space, ensuring scale-invariant and geometrically meaningful aggregation. BeVarMax, inspired by Otsu’s thresholding method, maximizes between-class variance to identify optimal groupings and uncover latent structure among alternatives, surpassing conventional clustering techniques such as k-means in robustness and global optimality. Using longitudinal data from 13 East Asia Summit (EAS) countries spanning 2012 to 2023, this model is applied to measure national road safety performance based on 15 tailored safety performance indicators (SPIs). Results demonstrate the model’s reliability, robustness, and superior discriminative power across normalization and weighting schemes, validated through extensive sensitivity and benchmarking analyses. Policy implications are twofold: it enables benchmarking of high and low performers to guide targeted interventions, and it supports strategic resource allocation by identifying priority areas such as enforcement, infrastructure, and behavioral factors. The proposed model serves as a practical decision-support tool for monitoring progress and fostering regional cooperation in line with global road safety goals.