Screening out hazardous corridors using project-level Empirical Bayes:Case Study: Khorasan Razavi Province

A mainstep towards safety management system is identifying hazardous road locations. This can be implemented in either project or site-specific level. Many situations especially like optimizing and allocating safety budgets within a jurisdiction; Prioritizing corridors for inspection, scheduling and distributing Police patrols and mobile emergency management and lack of enough precision in crash data over the segment level, application of project-level Empirical Bayes method would give an appropriate estimation of project-level(corridor) crashes in order to identify corridors with promise. The present study is an application and introduction of project-level EB method based on fatal and injury crashes of khorasan Razavi province using 1391&92 crash data. For this to achieve, safety performance functions with the aid of generalized linear model assuming negative binomial distribution for error term were separately developed for each severity level. For developing the models 69 homogeneous from 31 corridors of khorasan Razavi province were addressed. Then, through aggregating crash recorded data and estimated ones over the segments of each corridor using project-level EB method an appropriate estimation of 31 corridors were obtained; therefore, prioritizing was fulfilled. The results revealed that road type and vehicle-kilometer driven (VKT) had a significant effect on crashes across different injury levels. Evaluating the models according R2adj and cumulative residual plot (CSR) indicated that the models have satisfactory performance at 95% significant level. Eventually, Sabzevar-Kahak, Baghcheh-Torbat Heidarie and Sarakhas-Mashad were prioritized as the triplicate first ranks for further investigation.