Does Traffic Congestion Increase Vehicle Crashes? A Regional Scale Study with a Novel Congestion Index and Machine Learning Methods

Keywords: traffic congestion, vehicle crash, urban transportation, MileHour, Montgomery
Abstract Type: Virtual Paper Abstract

Abstract

Traffic congestion and vehicle crashes cause significant socioeconomic costs. Even if their causal relationships are frequently observable in the real world, a literature review reveals that little quantitative analyses have been performed about their relationships and their spatiotemporal patterns at a regional scale. One reason might be attributed to the lack of appropriate tools to measure the traffic data that covers a large area. With the MileHour congestion unit that was recently developed for measuring traffic amounts, this study tackled the relationships between traffic and crashes and their spatiotemporal patterns. Montgomery County, Maryland, was chosen as a case study area. Congestion information was collected from Google online traffic maps, and a spatiotemporal traffic database was constructed. Crash locations were also collected as a GIS layer with their occurrence times. Results show significant relationships between crashes and congestions, geographically and temporally. Spatial pattern analysis results also show spatially clustered crash and traffic patterns. With congestion amounts and spatiotemporal characteristics, a model was also developed using linear regression and machine learning methods.