TY - GEN
T1 - Severity Analysis of Heavy Vehicle Crashes Using Machine Learning Models
T2 - International Conference on Transportation and Development 2021: Transportation Operations, Technologies, and Safety, ICTD 2021
AU - Hasan, Ahmed Sajid
AU - Kabir, Md Asif Bin
AU - Jalayer, Mohammad
N1 - Publisher Copyright:
© 2021 International Conference on Transportation and Development 2021: Transportation Operations, Technologies, and Safety - Selected Papers from the International Conference on Transportation and Development 2021. All rights reserved.
PY - 2021
Y1 - 2021
N2 - Large trucks are a vital mode for freight transportation. Increasing demand in freight transportation increases the risk of truck-involved crashes on highways. Truck-involved crashes counted for 4,761 deaths in the United States during the year 2017, which is 12% more than the death toll of the year 2008. We gathered four years (2016-2019) of motor vehicle crashes involving large trucks in New Jersey for further analysis. The data set is classified into three injury severity categories, including severe injury, possible injury, and no injury. To predict the crash severity, we trained and tested the data set with the three most commonly used machine learning models, including support vector machine, random forest, and boosting methods. The performance of the models is evaluated by their precision, accuracy, and recall. Further, a sensitivity analysis was performed to demonstrate the impact of the top contributing factors in truck-related crashes. The findings of the study will help practitioners and policymakers determine the effects of influential parameters on injury severity outcomes and take necessary countermeasures to minimize the frequency and severity of large-truck crashes.
AB - Large trucks are a vital mode for freight transportation. Increasing demand in freight transportation increases the risk of truck-involved crashes on highways. Truck-involved crashes counted for 4,761 deaths in the United States during the year 2017, which is 12% more than the death toll of the year 2008. We gathered four years (2016-2019) of motor vehicle crashes involving large trucks in New Jersey for further analysis. The data set is classified into three injury severity categories, including severe injury, possible injury, and no injury. To predict the crash severity, we trained and tested the data set with the three most commonly used machine learning models, including support vector machine, random forest, and boosting methods. The performance of the models is evaluated by their precision, accuracy, and recall. Further, a sensitivity analysis was performed to demonstrate the impact of the top contributing factors in truck-related crashes. The findings of the study will help practitioners and policymakers determine the effects of influential parameters on injury severity outcomes and take necessary countermeasures to minimize the frequency and severity of large-truck crashes.
UR - http://www.scopus.com/inward/record.url?scp=85108028163&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85108028163&partnerID=8YFLogxK
U2 - 10.1061/9780784483534.025
DO - 10.1061/9780784483534.025
M3 - Conference contribution
AN - SCOPUS:85108028163
T3 - International Conference on Transportation and Development 2021: Transportation Operations, Technologies, and Safety - Selected Papers from the International Conference on Transportation and Development 2021
SP - 285
EP - 296
BT - International Conference on Transportation and Development 2021
A2 - Bhat, Chandra R.
PB - American Society of Civil Engineers (ASCE)
Y2 - 8 June 2021 through 10 June 2021
ER -