Rear-end collisions are the single most frequent type of crashes on freeways. Moreover, since almost all of them occur during periods of medium to heavy demand, their impact on traffic operation remains most noticeable. Using the traffic speed data from multiple locations surrounding the crash site rear-end crashes may be categorized into two clusters. These clusters are separable from each other based on simple 'if-then-else' rules formulated by classification tree methodology. Upon further analysis it was found that if a random sample of traffic speed data is subjected to these rules then one set of conditions occur more frequently than the conditions belonging to the other cluster. Upon further analysis of these two groups of crashes it was observed that the rear-end crashes which are observed under rare traffic conditions are slightly higher frequency of Injury crashes than the crashes that are observed in more prevalent traffic. The frequency patterns of these two groups of crashes are also examined with respect to parameters such as time of day, day of week and freeway location. (Author/publisher).
Abstract