Abstract

Road traffic accidents are one of the leading causes of global mortality, especially among young people and adults, and they have significant impacts on essential areas such as health and the economy. This study aims to compare and select a model to classify the causes of accidents on federal highways in Minas Gerais, Brazil, using data processing methods and Machine Learning techniques. Data from the Federal Highway Police, from January 1, 2023, to September 30, 2023, were used to analyze the Random Forest (RF), Support Vector Machine (SVM), and K-Nearest Neighbors (K-NN) algorithms. The SVM showed the best accuracy and Kappa index, while the RF had reasonable performance, and the K-NN was inferior and slower. The analysis reinforces the importance of carefully choosing the model, considering performance and computational efficiency. The study aims to support security authorities, such as the Federal Highway Police, in analyzing and recording occurrences, strengthening the construction of a robust and reliable database for filling out the Traffic Accident Report.

Keywords

• Supervised Learning
• K-Nearest Neighbor
• Random Forest
• Support Vector Machine

References

• DETRAN-MG. Registro Nacional de Acidentes e Estatísticas de Trânsito. 2024. Disponível em: https://transito.mg.gov.br/sobre1/estatisticas/registro-nacional-de-acidentes-de-transito. Acesso em: 15 jan. 2024.
• HOSSIN, M. N.; SULAIMAN A. Review on Evaluation Metrics for Data Classification Evaluations. International Journal Of Data Mining & Knowledge Management Process, [S.L.], v. 5, n. 2, p. 01-11, 31 mar. 2015. Academy and Industry Research Collaboration Center (AIRCC). http://dx.doi.org/10.5121/ijdkp.2015.5201
• LIAW A, WIENER M (2022). Random Forest: Breiman and Cutler's Random Forests for Classification and Regression. R package version 4.7-1.1. https://cran.r-project.org/web/packages/randomForest/index.html
• MALAQUIAS, E. O.; TOSTA, M. C. R.; CHAVES, G. L. D.; RIBEIRO, G. M. ACIDENTES EM RODOVIAS BRASILEIRAS: um estudo com técnicas de machine learning para classificar a causa das ocorrências. In: 35° CONGRESSO DE PESQUISA E ENSINO EM TRANSPORTE DA ANPET, 35., 2021, online. [S. I.]: Anpet, 2021. p. 2322-2334.
• MEGNIDIO-TCHOUKOUEGNO, M; ADEDEJI, J. A. Machine Learning for Road Traffic Accident Improvement and Environmental Resource Management in the Transportation Sector. Sustainability, [S.L.], v. 15, n. 3, p. 2014, 20 jan. 2023. MDPI AG. http://dx.doi.org/10.3390/su15032014
• MEYER, D.; DIMITRADOU, E.; HORNIK, K.; WEINGESSEL, A.; LEISCH, F. e1071: Misc Functions of the Department of Statistics, Probability Theory Group (Formerly: E1071), TU Wien. R package version 1.7-16, 2023. https://CRAN.R-project.org/package=e1071
• MINAS GERAIS. SECRETARIA DE SAÚDE DE MINAS GERAIS. Acidentes por Transporte Terrestre. Disponível em: http://vigilancia.saude.mg.gov.br/index.php/paineis-tematicos/. Acesso em: 15 jan. 2024.
• PRF (Polícia Rodoviária Federal). Dados Abertos da PRF. 2023. Disponível em: https://www.gov.br/prf/pt-br/acesso-a-informacao/dados-abertos/dados-abertos-da-prf. Acesso em: 12 jan. 2023.
• R CORE TEAM. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2023. ISBN 3-900051-07-0, URL http://www.R-project.org
• RAINIO, O.; TEUHO, J.; KLÉN, R. Evaluation metrics and statistical tests for machine learning. Scientific Reports, [S.L.], v. 14, n. 1, p. 255-269, 13 mar. 2024. Springer Science and Business Media LLC. http://dx.doi.org/10.1038/s41598-024-56706-x
• RIPLEY, B.; VENABLES, W. class: Functions for Classification. R package version 7.3-22, 2023. https://cran.r-project.org/web/packages/class/index.html
• World Health Organization. Despite notable progress, road safety remains urgent global issue. 2023. https://www.who.int/news/item/13-12-2023-despite-notable-progress-road-safety-remains-urgent-global-issue. Acesso: 10 jan. 2024.
• World Health Organization. Global status report on road safety 2023. 2023. Disponível em: https://www.who.int/teams/social-determinants-of-health/safety-and-mobility/global-status-report-on-road-safety-2023. Acesso em: 10 jan. 2024.
• World Health Organization. Road safety Brazil 2023 country profile. 2023. Disponível em: https://www.who.int/publications/m/item/road-safety-bra-2023-country-profile. Acesso em: 10 jan. 2024.
• XU, Haojie et al. PFD-Assisted Sampling PLL With Seamless PFD-SPD Switching Scheme and Supply-Insensitive RO. Ieee Microwave And Wireless Technology Letters, [S.L.], v. 33, n. 10, p. 1474-1477, out. 2023. Institute of Electrical and Electronics Engineers (IEEE). http://dx.doi.org/10.1109/lmwt.2023.3307733

Paper information