Modelos categóricos aplicados à análise exploratória dos níveis de agregação da severidade de acidentes compedestres emáreas urbanas.

Abstract

Promoting a safer road environment for pedestrians requires an understanding of the risk factors associated with the injuries suffered by these users when they are involved in crashes. This study is dedicated to investigating different severity classification configurations in the categorical modeling and identifying factors that influence the severity of crashes with pedestrians. For this purpose, multinomial logit models were estimated by using crashes records from the city of Fortaleza between the years 2017 and 2019. Variables related to pedestrians, aspects of the crash and characteristics of the urban structure were investigated. The results indicated that the model with three levels of severity (light/moderate, severe and fatal) showed the best performance. From this model, it was observed that factors such as pedestrian's advanced age, accidents at night, with the involvement of heavy vehicles, on weekends and located on roads with higher road classification, are associated with more severe injuries.