Investigating Factor Associations in Barrier Crashes through Cluster Correspondence Analysis

Roadside and median barriers have proven effective in preventing crashes; however, a significant number of crashes still occur that involve road barriers across United States. This study focuses on analyzing the factors related to barrier crashes across Texas. The dataset includes 63,475 crashes involving road barriers and covering six years of crash data (2017–2022). Using cluster correspondence analysis (CCA) data mining approach, the study identifies six different clusters of crashes. Each cluster identifies factors contributing to barrier-related crashes. Before implying the CCA, variable importance analysis is also performed to identify the significance of the variables. The analysis highlighted that dry road surface conditions and clear weather are highly associated with high-speed crashes. Driver distraction and absence of traffic control devices can attribute to the crashes on roads with lower speed limits. Moreover, along with other factors, adverse weather conditions are also found to be a contributing factor that can influence the crash frequency and type of crashes. The analysis also implies that non-complex crash type related to a single vehicle are highly correlated with barrier crashes. The study concludes by making several policy-making implications to assist the transportation planners in reducing the frequency and severity of barrier-related crashes.