1887
Proceedings of the 24th World International Traffic Medicine Association Congress, Qatar 2015
  • ISSN: 2223-0440
  • EISSN:

Abstract

This research considers a Bayesian analysis of crash data in an attempt to predict, from a group of potential collision hotspot sites, which of these sites could benefit from treatment with a road safety scheme. Intrinsic to the analysis is the identification of trend and site-specific regression to the mean (RTM) effects. As in a standard retrospective before-after study to evaluate the effectiveness of a change in e.g. the geometric design of an intersection, observed collision rates are adjusted using values from a suitable crash prediction model (CPM). In any year, collision rates, which are unusually high/low will be suitably depressed/inflated according to the posterior distributions for collision rates at each site, hence giving a more realistic summary of safety in that year. Where site characteristic information (e.g. annual figures for average speed or traffic flow) for use in the CPM is limited, standard techniques from time series analysis are employed to exploit any time dependent (autoregressive) structure observed in historical collision rates at each site. The Bayesian posterior predictive distribution is then used to predict collision rates at each site in future years, having adjusted for trend, RTM and any autoregression in collision rates at each site. This equips road safety practitioners with the necessary methodology to identify, and possibly treat, such locations before these collisions occur and this has the potential to help, inform, and direct, investment in future road safety schemes. In this research, crash data from the United Kingdom and Germany were analysed, and results have shown that this methodology is transferable between regions. The methodology is currently being implemented into a prototype software application to be tested by local road safety practitioners later in the year in the UK.

Loading

Article metrics loading...

/content/journals/10.5339/jlghs.2015.itma.36
2015-11-12
2020-10-30
Loading full text...

Full text loading...

http://instance.metastore.ingenta.com/content/journals/10.5339/jlghs.2015.itma.36
Loading
  • Article Type: Research Article
This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error