Published In
International Journal of Transportation Science and Technology
Document Type
Article
Publication Date
2020
Subjects
Transportation -- Safety, Traffic monitoring
Abstract
The high number of vehicle–pedestrian crashes in the United State has gained increased attention among transportation safety analysts in recent years. Being directly exposed to the collision force makes pedestrians more prone to becoming severely injured when in crash than other road users. Considering the fact that pedestrian-involved crashes is a serious public health problem, the current study’s aim is to investigate the contributing factors associated with injury severity of pedestrian crashes by time-of-week. Separate injury severity models for weekday and weekend crashes were developed, and the overall stability of the model estimates was examined through likelihood ratio tests. For this purpose, random parameter ordered-response models were employed to specify the ordinal nature of injury severity levels and capture the potential unobserved heterogeneity. In addition, Artificial Neural Network (ANN) was used to explore the nonlinear relationship between explanatory variables and severity outcomes. Comparison of the prediction performance demonstrated that optimized ANN provides superior results compared to conventional statistical approaches. A variable impact analysis was then conducted on the optimized ANN to investigate the effects of the explanatory variables on injury severity. The results revealed the factors that are significantly associated with pedestrian fatalities. These findings further provide insights for a better understanding of pedestrian injury severity in weekday vs. weekend crashes through the impact analysis of various explanatory variables.
Locate the Document
DOI
10.1016/j.ijtst.2020.01.001
Persistent Identifier
https://archives.pdx.edu/ds/psu/33170
Citation Details
Mokhtarimousavi, S., Anderson, J. C., Azizinamini, A., & Hadi, M. (2020). Factors affecting injury severity in vehicle-pedestrian crashes: A day-of-week analysis using random parameter ordered response models and Artificial Neural Networks. International Journal of Transportation Science and Technology.
Description
Copyright 2020 Tongji University and Tongji University Press. Publishing Services by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).