City-scale GPS data reveals impact of spatial configuration and dedicated infrastructure on e-scooter route choice

Abstract

Abstract Shared e-scooter use has rapidly expanded in major cities worldwide, offering promising solutions for sustainable transport and new data sources to advance the science of cities. This study leverages a city-scale GPS dataset of 14,029 e-scooter trips recorded over a three-month period in 2021 within the Mannheim/Ludwigshafen metropolitan area in Germany. For the first time, our analysis integrates the discrete choice modelling framework with space syntax theory using such large-scale revealed preference data, uncovering new insights into the impact of spatial configuration on routing behaviour. The results highlight the significant role of spatial configuration in e-scooter routing, with space syntax metrics consistently improving model performance and suggesting that riders avoid both places that are not well-integrated on a regional and highly accessible on a local level. Results also reveal that dedicated bicycle infrastructure, including bike lanes and tracks, reduces perceived travel distance by over 51% for e-scooter riders. Additionally, riders exhibit context-dependent behaviour, favouring pedestrian spaces during busy weekdays while avoiding them at other times. These insights can guide policymakers in designing micro-mobility-friendly urban environments.