Identifying Hotspots on Freeways using the Continuous Risk Profile with Hierarchical Clustering Analysis

Abstract

Crashes that occur on freeways generally cause extensive damage and injuries. Therefore, there is a need for the development of techniques for managing and reducing the number of crashes that occur by identifying hotspots efficiently within a limited budget. Among existing network screening methods, the Continuous Risk Profile(CRP) model well known to have performance that is superior to competing methodologies. However, to identify hotspots, the CRP model requires the use of safety performance functions which are used as a rescaling factor. In this study, I utilized hierarchical clustering analysis to use the Continuous Risk Profile, which had great results for identifying hotspots in nations and regions in which no safety performance functions have been established. I identified hotspots by replacing safety functions that are used as a rescaling factor in the CRP model with expected average crash frequency following groups that were obtained by hierarchical clustering analysis. I compared the hotspots identified by the existing CRP model and the hotspots identified by the CRP model using hierarchical clustering analysis. Also, I compared the hotspots identified by the CRP model using hierarchical clustering analysis and the Sliding Moving Window method and the Peak Searching method. These comparisons indicated that the CRP model using hierarchical clustering analysis, just like the existing CRP model, was more effective at identifying hotspots on freeways than other network screening methods.