Exploring the correlation between hard-braking events and traffic crashes in regional transportation networks: A geospatial perspective

Abstract

Traffic crashes are deemed a leading cause of death and injury in the United States. To improve traffic safety, historical traffic crash data are typically analyzed with a focus on factors such as roadway geometry and traffic volume. However, due to the infrequent and sporadic nature of traffic crashes, obtaining traffic safety evaluation for specific roadways requires a time and resource-intensive process, which involves extended periods of data collection and rigorous statistical reasoning. This paper explores alternative approaches, using hard-braking data collected from connected vehicles to develop a cost-efficient surrogate traffic safety measure. The geospatial correlations between hard-braking events and traffic crash locations are examined through two geospatial analysis methods: colocation analysis and network cross K-function. A case study was conducted in northern Nevada to identify hard-braking hot spots and reveal the overall cluster pattern. The colocation analysis identified that individual hard-braking events can be spatially related to crashes based on the network cross K-function result. The cases of four tracts in Reno, Nevada also demonstrate that the selection of clustering distances can influence the correlation between hard braking events and traffic crashes. This study shows the potential of using connected vehicle data to produce safety analyses for transportation networks.