Thesis
A two-stage model for predicting crash rate by severity types
Washington State University
Master of Science (MS), Washington State University
2018
Handle:
https://hdl.handle.net/2376/102252
Abstract
This research extends the prediction models that calculate the current crash counts by severity (CCS) by developing a two-stage regression model for formulating a new CCS-based hotspot identification (HSID) method. Specifically, this work studies the impacts of roadway geometry, weather condition, and traffic characteristics on the crash rate (number of crashes per mile) by severity on freeway facilities. In the proposed two-stage regression model, a logistic regression models the probability of different crash severity level at first and then, linear regression models estimate the crash rate by severity. The proposed model does not involve complex data structure and introduce duration based weight to account for the impacts inclement weather conditions. We fitted the model to a dataset including 828 road segments on all Interstate Highways in Washington State, USA during 2011-2014. Our findings indicated that the two-stage model outperformed a benchmark linear regression model in terms of method consistency, rank difference, and total score in HSID evaluation test and provided a better understanding of the causality of different severity types. Results showed that AADT per lane and the number of lanes had the most profound effects on crash rate. The proposed model revealed that asphalt paved shoulders, wider shoulder widths, and curved segment with wider outer shoulders are associated with lower crash rates.
Metrics
18 File views/ downloads
30 Record Views
Details
- Title
- A two-stage model for predicting crash rate by severity types
- Creators
- S. M. A. Bin al Islam
- Contributors
- Ali Hajbabaie (Degree Supervisor)
- Awarding Institution
- Washington State University
- Academic Unit
- Civil and Environmental Engineering, Department of
- Theses and Dissertations
- Master of Science (MS), Washington State University
- Publisher
- Washington State University; [Pullman, Washington] :
- Identifiers
- 99900525061501842
- Language
- English
- Resource Type
- Thesis