Leveraging passively generated big data to model e-scooter routing considering spatial configuration

E-scooters constitute a novel, technology-driven mode of transport that provides an opportunity to support the transport sector in becoming increasingly safe, environmentally friendly, and equitable while ensuring the economic viability of its operations. At the same time, the understanding of e-scooter riders’ behaviours as well as the appropriate utilisation of the rich passively generated data produced by the mode remains immature, with negative implications not only for their successful integration in the transport and urban planning ecosystem but also for the realisation of their potential to support developing as well as testing theories. One such theory is the theory of natural movement which posits that the configuration of space is the prime generator of movement. While it is well-examined for traditional modes of transport, its applicability to the movement of e-scooters, the relationship of spatial configuration with individuals’ movements, as well as its robustness and methodological soundness in light of large movement datasets are not well-investigated.

This thesis bridges these gaps by i) systematically reviewing the state-of-the-art of passively generated data in the micro-mobility field, including the identification of possible methodological, conceptual, and empirical advancements, ii) testing and expanding the theory of natural movement for the mode of e-scooters while taking advantage of a large passively generated e-scooter tracking dataset, and iii) developing a discrete route choice model for e-scooters which, as being based on passively generated data collected at a city scale, is not only the first of its kind but also utilised to understand the impact of spatial configuration on individuals’ routing decisions.

The findings of the thesis underline the advantages of passively generated data regarding coverage, resolution, analysis techniques, the capability to respond to new research questions, and avoiding human recording and inconsistency errors. The provided classification scheme for passively generated micro-mobility data supports the identification of research gaps and opportunities on the levels of data types, derived information, possible interventions, and identified challenges while highlighting the need for considering intrinsic bias, inaccuracies, and privacy concerns.

Furthermore, the thesis presents an extension of existing theory of natural movement-driven methods in utilising a spatial econometrics-based framework to account for the in comparison to past research high data density. It demonstrates that population density or points-of-interest iii

distribution have a limited explanatory effect on e-scooter flows when compared to spatial configuration parameters. The developed models are spatio-temporally cross-validated utilising e-scooter tracking data from three case studies and allow for the identification of hotspots of e-scooter usage in a new location

Moreover, the route choice modelling results presented in this thesis quantify the positive effect of cycling infrastructure on e-scooter riders’ route choices, suggesting that cycle lanes and tracks reduce perceived travel distance by at least 51%, and demonstrate that riders exhibit time-dependent behaviour, favouring pedestrian spaces during busy weekdays while avoiding them at other times. Additionally, the thesis’s findings suggest that spatial configuration cannot only be used to model traffic flows but that it also plays a role in individuals’ route choices as the inclusion of spatial configuration parameters significantly increases model fit.

The results are relevant to research and practice, offering direct support to policymakers and planners who aim to create safe, environmentally sustainable, economically viable, and equitable transport systems. They can use the findings to inform data collection and analysis strategies, regulations that meet the needs of e-scooter riders while not curtailing others, designing infrastructure and allocate street space in a way that benefits micro-mobility as a conglomerate rather than separately, as well as building urban morphologies that incentivise sustainable travel.