Signalized Intersection Hotspot Identification Based on Potential for Safety Improvement

Abstract:

The Bayesian based potential for safety improvement method identifies signalized intersection hotspots using the difference between the Bayesian posteriors of crash number and the expected crash number for similar sites which are obtained from the regression model. As the difference increases, the potential for safety improvement increases. In addition, the Bayesian method combines clues from both the observed crash frequency of a specific site and the expected crash number for similar sites to estimate the crash number which can overcome the problem associated with the fluctuation of observed crashes. Based on signalized crash data at intersections in Shanghai, the full Bayesian based potential for safety improvement method is compared with the crash frequency method, the empirical Bayesian method, the full Bayes method and the empirical Bayesian based potential for safety improvement method. The results show that the proposed full Bayesian potential for safety improvement is superior to other methods.