Safety Investigation of Nonmotorized Crashes in the City of Huntsville, Alabama, Using Count Regression Models

While walking and cycling can be enjoyable, there are potential safety risks associated with these modes, especially when interacting with automobiles. This study contributes to the safety of non-motorized transportation by applying and comparing three zero-inflated count models to a collection of both bike and pedestrian crashes. The models studied are two variations of the zero - inflated negative binomial (ZINB) and the zero - inflated Poisson (ZIP). The ZINB was examined using the standard ZINB and a modified one that included a different variance estimator. In this study, sociodemographic (e.g., school enrollment, number of households and employment values), traffic (e.g., traffic volume and speed limit), and infrastructure or built environment variables (e.g., sidewalk length) were used. Five - year bike and pedestrian crashes (2010 through 2014) in 368 traffic analysis zones (TAZs) in the city of Huntsville, Alabama were used. The performance of the six fitted count models was compared based on the prediction performance and goodness - of - fit statistics. The prediction accuracy included the mean absolute deviance (MAD) and the mean square prediction error (MSPE). The standard ZINB outperformed the corresponding ZINB type and the ZIP for both bike and pedestrian crashes (especially in the relatively higher MAD and MSPE estimates, which represents higher prediction performance). The fitted regression models for both bike and pedestrian crashes in Huntsville showed that there was an increase in crashes with the increase in traffic volume, number of households, and amount of retail employment. The results of the fitted count models are deemed useful for decision makers to identify and predict high - risk zones for bicyclists and pedestrian crashes in a city or county.

• Supplemental Notes:
  • This paper was sponsored by TRB committee ANB20 Standing Committee on Safety Data, Analysis and Evaluation.
• Authors:
  • Shirani, Niloufar
  • Doustmohammadi, Mehrnaz
  • Haleem, Kirolos
  • Anderson, Michael
• Conference:
  • Transportation Research Board 97th Annual Meeting
  • Location: Washington DC, United States
  • Date: 2018-1-7 to 2018-1-11
• Date: 2018

Language

• English

Media Info

• Media Type: Digital/other
• Features: Figures; Maps; References; Tables;
• Pagination: 17p

Subject/Index Terms

• TRT Terms: Bicycle crashes; Crash characteristics; Nonmotorized transportation; Pedestrian vehicle crashes; Risk assessment; Traffic analysis zones; Traffic volume
• Geographic Terms: Huntsville (Alabama)
• Subject Areas: Data and Information Technology; Pedestrians and Bicyclists; Safety and Human Factors;

Filing Info

• Accession Number: 01659002
• Record Type: Publication
• Report/Paper Numbers: 18-03065
• Files: TRIS, TRB, ATRI
• Created Date: Feb 5 2018 11:26AM