A Dynamic Least-Cost Path Method for Incorporating the Street-level Built Environment into Mode Choice Utility

Abstract

The street-level built environment (BE) describes the micro-environment we experience along paths and streets, such as greenness, slope, walking/cycling infrastructure, and motor vehicle traffic. With greater availability of street-level BE data, there is an emerging body of empirical literature linking street-level BE with mode choice. A common method for developing BE predictors for mode choice is to aggregate attributes along estimated routes between the trip origins and destinations. However, the requirement to pre-specify routing parameters has methodological and behavioural inconsistencies that could cause an underestimation of the significance and influence of street-level BE. This study proposes a method in which routing parameters adapt dynamically to the estimated BE predictors during maximum likelihood estimation. With a demonstration for Greater Manchester, we show that this method produces plausible outputs that can more effectively capture the influence of the street-level BE on behaviour.