Profile
IDENTIFICATION OF KEY RISK FACTORS FOR TRAFFIC ACCIDENTS USING MACHINE LEARNING
DOI: xxxxxxxx
Submitted: 27/11/2025
Accepted: 05/02/2026
Published: 06/02/2026

Abstract

Road safety represents a significant challenge for the transport sector due to the severe consequences of traffic accidents and the need for timely identification of factors that increase risk. This study aims to quantify the influence of selected parameters on the weighted accident index and to establish a foundation for predictive models capable of identifying high-risk road sections. The analysis covers 161 sections of the national road network in the Republic of North Macedonia, with a total length of approximately 1,300 km, and includes 23 parameters. The assessment was performed using machine learning techniques, with model evaluation conducted on an 80/20 train–test split. The results reveal that road characteristics and traffic volume (AADT) exert the greatest influence on accident risk, whereas environmental factors have minimal impact. This approach enables more efficient planning of interventions and contributes to the overall improvement of road safety.

Authors

  • Riste Ristov, Affiliation: Ss. Cyril and Methodius University in Skopje, Faculty of Civil Engineering Skopje , ORCID: 0000-0002-4996-0382

Keywords

  • accidents
  • machine learning
  • roads
  • safety
  • weighted index

Attachments

References

  • [1] World Health Organization. (2024). Road traffic injuries.
  • Available at: https://www.who.int/news-room/fact-sheets/detail/road-traffic-injuries (accessed March 2025).
  • [2] European Commission. (2023). Road Safety Statistics 2023. Available at: https://road-safety.transport.ec.europa.eu/european-road-safety-observatory_en (accessed March 2025).
  • [3] State Statistical Office of the Republic of North Macedonia. (2025).
  • MakStat – Statistical database. Available at: https://makstat.stat.gov.mk/PXWeb/pxweb/mk/ (accessed March 2025).
  • [4] Wang, Y., Zhang, Y., Wu, J., & Xu, C. (2023). Analyzing the Risk Factors of Traffic Accident Severity Using Machine Learning and Association Rules. International Journal of Environmental Research and Public Health, 20(1), 345. https://doi.org/10.3390/ijerph20010345
  • [5] Jiang, Y., Li, S., Zhao, Z., & Chen, H. (2024). Machine Learning-Based Prediction Analysis of Potential Factors Influencing Traffic Accident Severity. Sustainability, 16(2), 1125. https://doi.org/10.3390/su16021125
  • [6] Çelik, E., & Sevli, D. (2022). Predicting Road Traffic Accident Severity Using Machine Learning Techniques. Applied Sciences, 12(15), 7485. https://doi.org/10.3390/app12157485
  • [7] Public Enterprise for State Roads. Web-GIS platform for spatial analysis and visualisation. Available at: http://62.77.137.99/pesr/webgis/#/map (accessed March 2025).
  • [8] Ministry of Local Self-Government. (2021). Programme for Development of the Planning Regions for the period 2021–2026. Government of the Republic of North Macedonia, Skopje.
  • [9] Doncheva, R., Ognjenovic, S. (2024). Road Design. Ss. Cyril and Methodius University – Faculty of Civil Engineering, Skopje. ISBN: 978-608-4510-60-4.
  • [10] Tobias, P., de León Izeppi, E., Flintsch, G., Katicha, S., McCarthy, R. (2023). Pavement Friction for Road Safety: Primer on Friction Measurement and Management Methods. Federal Highway Administration (FHWA), Report No. FHWA-SA-23-007.
  • [11] Public Enterprise for State Roads. Web-GIS platform for spatial analysis and visualisation. Available at: http://tdps.roads.org.mk/ (accessed March 2025).
  • [12] Gjeshovska, V., Taseski, G., Ilioski, B. (2024). Intense precipitation in the Republic of North Macedonia. Ss. Cyril and Methodius University – Faculty of Civil Engineering, Skopje. ISBN: 978-608-4510-56-7.
  • [13] Government of the Republic of North Macedonia, Ministry of Transport, Project Unit. (2024). Black Spot Management Handbook (BSM). Skopje, July 2024.
  • [14] Patil, P., Du, J.-H., Kuchibhotla, A. K. (2022). Bagging in overparameterized learning: Risk characterization and risk monotonization. arXiv preprint,arXiv:2210.11445.https://doi.org/10.48550/arXiv.2210.11445
  • [15] Ke, G., Meng, Q., Finley, T. et al. (2017). LightGBM: A Highly Efficient Gradient Boosting Decision Tree. Advances in Neural Information Processing Systems, 30. Available at: https://proceedings.neurips.cc/paper_files/paper/2017/file/6449f44a102fde848669bdd9eb6b76fa-Paper.pdf
  • [16] Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A. V., & Gulin, A. (2018). CatBoost: unbiased boosting with categorical features. Advances in Neural Information Processing Systems,31.https://doi.org/10.48550/arXiv.1706.09516