Research context
Official crash records undercount cycling incidents, particularly minor single-bicycle crashes. A cyclist-safety indicator based only on recorded crashes therefore misses risk on many network links.
This thesis combines crash data with traffic conditions, infrastructure type, land use and activity levels, and cyclists' perceived safety to develop a more complete link-level indicator for Amsterdam.
Research questions
- How can crash, traffic, infrastructure, land-use, activity, and perceived-safety data be fused into a single link-level indicator?
- How strongly does each factor influence perceived safety, and how well does the resulting indicator perform against an established framework and on unseen data?
What you will do
- Review cyclist-safety indicators and methods for multi-source data fusion.
- Construct a link-level empirical model combining crash, traffic, infrastructure, land-use or activity, and perceived-safety data.
- Design and conduct a video-based survey to elicit perceived safety at link level.
- Validate the indicator against an established framework such as Level of Traffic Stress and through out-of-sample testing.
Background
The project suits a student with a strong interest in transport safety and urban cycling who is willing to work with multiple data sources and quantitative methods. Knowledge of regression modelling and survey design is required; support will be available for learning the required methods.
Project setting
The work is based on empirical data collected in Amsterdam during the thesis period within the PROSPER research programme at ORBIT Lab, which will support data access and collection.