Integrating machine learning and global sensitivity analysis for human-centered pedestrian path design

Abstract

This study examines the integration of human behavior into pedestrian path design, adopting a multidimensional approach that encompasses physical, psychological, social, environmental, and economic aspects. Recognizing a significant gap in current literature, particularly the limited consideration of integrated modeling frameworks and machine learning in developing urban contexts, this study aims to identify critical elements influencing pedestrian satisfaction and propose comprehensive, user-centered solutions. Theoretical foundations include behavioral science, urban planning, and architectural insights, while the empirical component involves data collection through direct observations, structured questionnaires, and personal interviews from users of a selected pedestrian path in New Asyut City, Egypt. A key methodological contribution of this research is the innovative application of Deep Residual Neural Networks (DRNNs) and Variance-Based Global Sensitivity Analysis (VBSA), allowing robust quantification and prioritization of design factors. Findings highlight the primary influence of easy access, safety, visual coherence, and climatic comfort, alongside significant interactions among these elements. Moreover, the study underscores critical shortcomings such as inadequate public transportation integration, inconsistent visual aesthetics, limited shading, and lack of economic activities along the pathway. By addressing these deficiencies and interactions comprehensively, this research provides urban planners and policymakers with a replicable analytical model and actionable insights to enhance pedestrian infrastructure. Also, the provided analysis framework is general and can be applied in other urban design contexts.