Urban Bus Demand Forecast at Stop Level: Space Syntax and Other Built Environment Factors. Evidence from Madrid

Resumen

Urban buses propose a challenge for traditional four-steps models of ridership estimation, as they require a different, closer scale approach, including the consideration of multiple possible stop-choices by travelers within walking distance. Thus, any model based on zoning and the bias of associating population to the nearest stop does not seem coherent in the case of urban bus. This study empirically examines the potential of possible ?attraction? descriptors, such as spatial integration (as described by the Space Syntax methodology) and other urban environment factors in order to estimate urban buses ridership by a direct forecast model based on multiple linear regression. Common explanatory factors found in the literature include population and employment in the vicinity area, as well as transport system service and performance. Some authors have claimed the predictive power of built environment variables (summarized by Cervero and Kockelman's three Ds: Density, Diversity and Design), which are supposed to describe pedestrian accessibility and attractiveness. This paper proposes that spatial-configurational measures (e.g. Space Syntax) could play an important role, given that these factors have proved themselves synthetic proxies for many urban processes and in order to describe spatial-configurational hierarchy and consequent attraction power. A demand forecast model at a stop level is explored by means of multiple linear regressions. Bus transport ridership at 84 stops in Madrid is forecasted using urban environment and spatial integration variables, as well as transport network accessibility indicators. Results seem encouraging and support that Space Syntax and other network integration variables could be an important asset for urban bus demand forecast models at a station level.