M.Sc Thesis

M.Sc StudentSultan Jody
SubjectExtracting spatial patterns of bicycle rides from
crowdsourced data
DepartmentDepartment of Civil and Environmental Engineering
Supervisor DR. Sagi Dalyot
Full Thesis textFull thesis text - English Version


Much is done nowadays to provide cyclists with safe and sustainable road infrastructure. New designated cycle ways are built to encourage citizens to use bicycles in urban environments and to decrease their interaction with pedestrians and vehicles. Still, the increasing traffic volumes cause accidents in shared spaces, such that investigation is still required to identify and better understand areas of high risks, resulting in the improvement of existing cycle way infrastructure. Still, a problem exist that not all relevant information can be gathered with data surveys and questionnaires: such a process is time consuming, costly, and is often limited in the number of citizens, locations, and travel activities studied. Such that the outcome produced by such efforts is mostly limited in understanding the individual motivations and limitations, and the characteristics of the built environment. This promotes the study of alternative practical means of data gathering. One of these alternatives is to use user-generated data, which is available today via online platforms of user-driven mapping projects, filling the gap of the aforementioned commonly used observation techniques, having the capacity of a more detailed and localized data collection.

User-generated content coming from regular citizen who voluntarily contribute data, information, or media, and the increasing pervasiveness of location-acquisition and communication technologies can help discovering usable knowledge about movement behaviour in urban environments. The content gives a glimpse into real data from other people without being biased by regular media. Still, some limitations exist with this type of data (in comparison to authoritative), e.g., accuracy, heterogeneity, completeness, such that tailored processes are required to make use of the data.

The hypothesis of this research is that an investigation of user-generated data that is collected voluntarily by cyclists in urban environments can reveal important and interesting spatial patterns and travel behaviours of cyclists, that otherwise could not have been retrieved and identified. To achieve this, data filtering and processing algorithms are developed to overcome inconsistencies in trajectories, and data noise and outliers. Map matching is used to analyse and mine the use of cyclists of the road network, also statistically analysing usage patterns to understand cyclists’ choices in urban environments.

Implementing the methodology developed here, showed that environmental factors, such as the attractiveness of green spaces and social areas, attracted cyclists in urban areas, preferring these over shorter direct routes. It was found that in the city of Amsterdam, having well established bicycle infrastructure and cycling culture, half of the cyclist usage are on designated cycle ways; this is different from Osnabrück, a city with a more limited bicycle infrastructure, having less than ten percent. In both cities, though not common, analysis showed cyclists still navigate through prohibited roads, a factor city planners can use for improving future planning. Moreover, this information, that perhaps cannot be mined and retrieved otherwise, can be used for the construction of a navigation application dedicated for cyclists, allowing them to commute in the city based on cyclers accumulated knowledge, choosing the path most appropriate for their preferences.