Methodology for Analyzing the Impacts of Highway Engineering Interventions over Traffic Safety Using GIS and Multinomial Logistic Regression

Abstract

Since the 1950s, Brazil has prioritized road transport for moving goods and passengers. Consequently, the country’s active vehicle fleet reached 65.8 million vehicles, and 96,366 crashes were recorded in 2016. This study presents a methodology for assessing the impact of road improvement projects on traffic safety using a geographic information system (GIS) and the R programming language. The analysis uses data provided by the Brazilian Federal Highway Police and comprises generating kernel density maps, assessing the spatial randomness of crashes using the quadrat count method, and fitting a multinomial logistic regression model to identify factors associated with crash severity. The methodology was applied to a 2.2 km road section in Niterói, Rio de Janeiro, where a road-widening project was completed in 2015. The results identified three crash hotspots and showed that crashes were not randomly distributed in space during either the pre- or post-intervention period. The multinomial logistic regression further indicated that crashes occurring after the intervention had higher odds of being classified in more severe outcome categories than crashes occurring before the intervention.