Machine learning (ML) techniques play a crucial role in today's modern world. Over the last years, road traffic safety is one of the applications where ML-methods have been successfully employed to prevent road users from being killed or seriously injured. A reliable data-driven predictive model is essential for this purpose. This could be achieved by successfully applying an intelligent transportation system to identify a driver at a higher risk of crashes. This study investigates the capabilities of different deep learning techniques to predict motorcycle crash severity. This study is based on 2,430 motorcycle crashes in a mountainous area in the United States over a 10-year period. Different deep networks (DNNs), including deep belief network, standard recurrent neural network (RNN), multilayer neural network, and single-layer neural network, were considered and compared in terms of prediction accuracy of motorcycle crash severity. Before conducting any analysis, feature reduction was performed to identify the optimal number of variables to include in the models by minimizing the error rate. Different metrics including the area under the curve and confusion matrix were used to compare the different models. Although the analyses were conducted on a relatively small dataset, the results indicate that almost all the DNN models better perform in predicting the severity of motorcycle crashes, compared with the single layer neural network. Finally, the RNN outperforms the other three neural network models. A comprehensive discussion has been made about the methodological approach implemented in this study.
Indian reservations hold the highest rate of crashes that lead to fatal and incapacitating injuries across the United States. Limited resources, absence of coordination across jurisdictions, and limited crash data make it difficult for tribes to reduce the number of these fatal and serious crashes. A safety toolkit was developed in this work to identify the high-risk crash locations and determine low-cost safety improvement countermeasures. The toolkit serves as a guideline and provides information, field examples, and resources in key topic areas to lead the effort to improve safety via the use of a five-step methodology. These five steps are, namely, compile data and crash data analysis, Level I field evaluation, combined ranking, Level II field evaluation to identify countermeasures, and benefit–cost analysis. The objective of the toolkit is to assist tribes to reduce the number of fatal and serious crashes, and provides flexibility for the tribes to utilize the methodology compatible with the onsite data, preferences of the tribes, and other factors to meet demands. The methodology described was successfully implemented on the Wind River Indian Reservation, Standing Rock Sioux Tribe, Sisseton Wahpeton Oyate Tribe, and the Yankton Sioux Tribe and showed great success in identifying high-crash locations.
Background: Historically, Indian reservations have been struggling with higher crash rates than the rest of the United States. In an effort to improve roadway safety in these areas, different agencies are working to address this disparity. For any safety improvement program, identifying high risk crash locations is the first step to determine contributing factors of crashes and select corresponding countermeasures. Methods: This study proposes an approach to determine crash-prone areas using Geographic Information System (GIS) techniques through creating crash severity maps and Network Kernel Density Estimation (NetKDE). These two maps were assessed to determine the high-risk road segments having a high crash rate, and high injury severity. However, since the statistical significance of the hotspots cannot be evaluated in NetKDE, this study employed Getis-Ord Gi* (d) statistics to ascertain statistically significant crash hotspots. Finally, maps generated through these two methods were assessed to determine statistically significant high-risk road segments. Moreover, temporal analysis of the crash pattern was performed using spider graphs to explore the variance throughout the day. Results: Within the Fort Peck Indian Reservation, some parts of the US highway 13, BIA Route 1, and US highway 2 are among the many segments being identified as high-risk road segments in this analysis. Also, although some residential roads have PDO crashes, they have been detected as high priority areas due to high crash occurrence. The temporal analysis revealed that crash patterns were almost similar on the weekdays reaching the peak at traffic peak hours, but during the weekend, crashes mostly occurred at midnight. Conclusion: The study would provide tribes with the tool to identify locations demanding immediate safety concerns. This study can be used as a template for other tribes to perform spatial and temporal analysis of the crash patterns to identify high risk crash locations on their roadways.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.