Skip to main content

Joe Matthews

Early detection of collision hotspots.

Email: [email protected]

Supervisors

Project description

Improving road safety is a global challenge. Road safety practitioners are becoming increasingly reliant on data to find solutions. Problems can arise in cases where data are limited. In such cases, we must account for confounding factors such as Regression To the Mean (RTM) to avoid misleading results.

We are investigating methods for road safety hotspot predictions. We are developing a model which proactively identifies future hotspots, enabling early treatment.

Our model combines network-wide covariate data and allows for local site effects and trends. We extend this to include seasonal data. We then fit structural models to the seasonal and spatial effects to reduce uncertainty in estimates. This also allows for interpolation between sites.

We add components to estimate collision severity and/or collision type factors. This again models network-wide and local effects. Thus, it allows for proactive detection of collision hotspots.

Publications

Qualifications

  • MMathStat, Newcastle University