ABSTRACT
The severity of roadway traffic crashes is a significant concern for highway safety officials around the world. Studying crash severity levels and contributing risk factors could lead to considerable savings in human life and economic losses. The study's main objective is to utilize a well-known machine learning technique, namely Random Forest, to analyze the severity levels of Sharjah's roadway traffic crashes. The study will identify the most critical exploratory factors among 23 roadway, driver, and environmental factors contributing to the crash's severity in terms of injury level. Towards this, a crash dataset for the 2015-2019 period was obtained from Sharjah Police, which included a total of 2,538 crash records, divided into three crash severity levels: fatal, hospitalized injury, and not-hospitalized injury. Several performance measures were computed using the h2o machine learning tool to find out the best-performing model, including confusion matrix, precision, recall, F1-measurement, among others. The dataset was split into a 70% training set and a 30% testing set. While the training set was used to develop the various models, the testing set was used to validate each model. Finally, the developed models were optimized by fine-tuning their parameters. The best-performing RF models achieved a classification accuracy for the training and test dataset is 61.9% and 58.2%, respectively, which is acceptable for classifying crash severity levels. The results revealed that the most significant factors associated with crash severity level in Sharjah were month of the year, collision type, day of the week, vehicle occupant, and nationality.
- M. F. Aljarrah, M. A. Khasawneh, and A. A. Al-Omari, "Investigating key factors influencing the severity of drivers injuries in car crashes using supervised machine learning techniques," J. Eng. Sci. Technol. Rev., vol. 12, no. 4, pp. 15–27, 2019, doi: 10.25103/jestr.124.03.Google ScholarCross Ref
- A. Das, A. Ghasemzadeh, and M. M. Ahmed, "Analyzing the effect of fog weather conditions on driver lane-keeping performance using the SHRP2 naturalistic driving study data," J. Safety Res., vol. 68, pp. 71–80, 2019, doi: 10.1016/j.jsr.2018.12.015.Google ScholarCross Ref
- R. Garrido, A. Bastos, A. De Almeida, and J. P. Elvas, “Prediction of road accident severity using the ordered probit model,” Transp. Res. Procedia, vol. 3, no. July, pp. 214–223, 2014, doi: 10.1016/j.trpro.2014.10.107.Google ScholarCross Ref
- A. Ahmadi, A. Jahangiri, V. Berardi, and S. G. Machiani, "Crash severity analysis of rear-end crashes in California using statistical and machine learning classification methods," J. Transp. Saf. Secur., vol. 12, no. 4, pp. 522–546, 2020, doi: 10.1080/19439962.2018.1505793.Google ScholarCross Ref
- I. Hammad, K. El-Sankary, and J. Gu, "A comparative study on machine learning algorithms for the control of a wall following robot," IEEE Int. Conf. Robot. Biomimetics, ROBIO 2019, pp. 2995–3000, 2019, doi: 10.1109/ROBIO49542.2019.8961836.Google ScholarDigital Library
- S. Mokhtarimousavi, J. C. Anderson, A. Azizinamini, and M. Hadi, "Improved Support Vector Machine Models for Work Zone Crash Injury Severity Prediction and Analysis," Transp. Res. Rec., vol. 2673, no. 11, pp. 680–692, 2019, doi: 10.1177/0361198119845899.Google ScholarCross Ref
- C. Xin , "Development of crash modification factors of horizontal curve design features for single-motorcycle crashes on rural two-lane highways: A matched case-control study," Accid. Anal. Prev., vol. 123, no. March 2018, pp. 51–59, 2019, doi: 10.1016/j.aap.2018.11.008.Google ScholarCross Ref
- C. Chen, G. Zhang, J. Yang, J. C. Milton, and A. Dely, "An explanatory analysis of driver injury severity in rear-end crashes using a decision table / Naïve Bayes ( DTNB ) hybrid classifier," Accid. Anal. Prev., vol. 90, pp. 95–107, 2016, doi: 10.1016/j.aap.2016.02.002.Google ScholarCross Ref
- A. Nickkar, A. Yazdizadeh, and Y. J. Lee, "Investigating factors that contribute to freeway crash severity using machine learning," Adv. Transp. Stud., vol. 52, pp. 131–142, 2020, doi: 10.4399/97888255370319.Google ScholarCross Ref
- D. Van Dao, A. Jaafari, M. Bayat, D. Mafi-gholami, and C. Qi, "Catena A spatially explicit deep learning neural network model for the prediction of landslide susceptibility," Catena, vol. 188, no. November 2019, p. 104451, 2020, doi: 10.1016/j.catena.2019.104451.Google ScholarCross Ref
- [11] K. Hamad, R. Al-Ruzouq, W. Zeiada, S. Abu Dabous, and M. A. Khalil, "Predicting incident duration using random forests," Transp. A Transp. Sci., vol. 16, no. 3, pp. 1269–1293, Jan. 2020, doi: 10.1080/23249935.2020.1733132.Google ScholarCross Ref
- L. Wahab and H. Jiang, "A comparative study on machine learning based algorithms for prediction of motorcycle crash severity," PLoS One, vol. 14, no. 4, pp. 1–18, 2019, doi: 10.1371/journal.pone.0214966.Google ScholarCross Ref
- C. Chen, G. Zhang, R. Tarefder, J. Ma, H. Wei, and H. Guan, "A multinomial logit model-Bayesian network hybrid approach for driver injury severity analyses in rear-end crashes," Accid. Anal. Prev., vol. 80, pp. 76–88, 2015, doi: 10.1016/j.aap.2015.03.036.Google ScholarCross Ref
- M. Taamneh, S. Alkheder, and S. Taamneh, "Data-mining techniques for traffic accident modeling and prediction in the United Arab Emirates," J. Transp. Saf. Secur., vol. 9, no. 2, pp. 146–166, 2017, doi: 10.1080/19439962.2016.1152338.Google ScholarCross Ref
- J. Ou, J. Xia, Y. J. Wu, and W. Rao, "Short-term traffic flow forecasting for urban roads using data-driven feature selection strategy and bias-corrected random forests," Transp. Res. Rec., vol. 2645, no. 1, pp. 157-167., 2017, doi: 10.3141/2645-17.Google ScholarCross Ref
Index Terms
- Severity Analysis of Sharjah Crashes Using Random Forest
Recommendations
Modeling Injury Severity of Vehicular Traffic Crashes
ICACR 2017: Proceedings of the 2017 International Conference on Automation, Control and RobotsData mining techniques constitute an alternative approach that has received increasing attention from researchers in recent years in road safety analysis field. In this paper Bayesian networks were used to develop models in order to identify factors ...
Crash analysis and evaluation of vehicular impacts on W-beam guardrails placed behind curbs using finite element simulations
Developed FE models of single-faced W-beam guardrails installed behind curbs.Performed nonlinear FE simulations of vehicles impacting roadside curb-guardrails.Evaluated the effects of a curb on the performance of W-beam guardrails at different placement ...
IoT-enabled highway safety pre warning system
IoTDI '19: Proceedings of the International Conference on Internet of Things Design and ImplementationWe propose to demo an IoT-enabled highway safety pre warning system which can inform the drivers of the danger ahead. In this system, we use road side unit (RSU) installed on highway guardrail to collect motion information from the on board unit (OBU) ...
Comments