Integrating machine learning and global sensitivity analysis for modeling public transport acceptance: evidence from Egyptian cities

This study investigates the factors influencing public transport acceptance in Egyptian cities using advanced machine learning techniques, including Optimal Regression Forest (ORF) and Variance-Based Sensitivity Analysis (VBSA). We analyze data from a survey of 2,511 respondents using a stated preference experiment to identify key drivers of acceptance and quantify their impact. VBSA was applied, quantifying both direct and interaction effects of fourteen explanatory variables. The results show that socioeconomic variables, perceived benefits, and service reliability are the most influential factors, with significant implications for policymakers seeking to improve public transport adoption. By integrating ORF and VBSA, our model achieves strong predictive performance, providing actionable insights for enhancing transport policy and service design. These findings contribute to the literature on public transport planning and offer practical recommendations for urban planners and policymakers in Egypt and in other Middle East and North Africa (MENA) cities with comparable socioeconomic and institutional conditions.