Pattern recognition from injury severity types of frontage roadway crashes

Abstract Frontage roads are the supporting roadways that are along freeways and fully controlled principal arterial roadway networks in the U.S. These roads are designed in a way to provide access between the freeways, principal arterials, and surrounding business entities. For Texas, these roadways are the leading design resolution for providing access along rural freeways and principal arterial roadways. These roadways are generally two-ways for rural and less developed urban areas and are mostly one-way for urban and city-centered roadways. Although frontage roadways possess major safety concerns, the safety performance of these roadways has not been well studied. This study collected six years of frontage road crash data from Texas to determine the patterns of associated factors by applying a dimension reduction method known as cluster correspondence analysis (CCA). The results revealed four clusters for each of the two datasets based on crash injury types. For fatal and injury crashes, the major clusters are distraction-related crashes at signalized intersections, segment-related crashes at dark unlighted conditions, yield signed intersection locations and segments with no TCDs, and intersection crashes on undivided roadways. For the no injury crash dataset, the key clusters are segment crashes in dark conditions and rain, crashes at signalized intersections with both drivers going straight, segment crashes with both drivers going straight with marked lanes or no TCDs, and intersection-related collisions on undivided roadways. Based on the evaluation results, suitable safety countermeasures and policy initiatives to reduce frontage road crash frequencies can be singled out.