Previsão de roubos de veículos no Rio de Janeiro utilizando algoritmos de Machine Learning

Gustavo de Souza Coelho; Lisleandra Machado; Domingos Sávio da Cunha Garcia; Leonardo Amorim de Araújo

doi:10.23925/ddem.v.3.n.15.72333

Authors

Gustavo de Souza Coelho Instituto Federal do Sudeste de Minas Gerais, IF SUDESTE, MG https://orcid.org/0009-0006-3113-3609
Lisleandra Machado Instituto Federal do Sudeste de Minas Gerais, IF SUDESTE, MG https://orcid.org/0000-0002-7761-8023
Domingos Sávio da Cunha Garcia Universidade Estadual de Campinas - UNICAMP, Campinas, SP https://orcid.org/0000-0002-8754-6780
Leonardo Amorim de Araújo Universidade Federal do Rio de Janeiro - UFRJ - Rio de Janeiro, RJ https://orcid.org/0000-0003-2722-7531

DOI:

https://doi.org/10.23925/ddem.v.3.n.15.72333

Keywords:

neural networks, decision trees, predictive models, Public Security, Vehicle Theft

Abstract

Predictions of crime variables are useful for tactical resource deployment, personnel allocation, and strategic planning, increasing the tactical and strategic awareness of police forces. In the United States, predictive methods are widely used to anticipate trends and identify locations and individuals prone to future victimization. However, national literature still lacks models capable of anticipating crime trends, with some exceptions, such as the CrimeVis tool and other studies with more modest prediction techniques. The evolution and popularization of artificial intelligence (AI), especially machine learning, allows for faster, cheaper, and more accurate predictions, helping to solve problems earlier. The application of AI can process and analyze large amounts of information quickly, improving decision-making. The main objective of this work is to evaluate which modeling approach – decision trees or neural networks – offers greater accuracy in predicting vehicle thefts in the city of Rio de Janeiro, contributing to the development of predictive tools that can assist authorities and public security agencies in the efficient allocation of resources and in the strategic planning of prevention policies. The accuracy of the models will be compared using the Mean Squared Error (MSE).

Author Biographies

Gustavo de Souza Coelho, Instituto Federal do Sudeste de Minas Gerais, IF SUDESTE, MG

Bachelor of Science in Computer Science.

Lisleandra Machado, Instituto Federal do Sudeste de Minas Gerais, IF SUDESTE, MG

Professor and Researcher funded by CNPQ, FAPEMIG, FUNDEP, and CAPES. Holds degrees in Law, Business Administration, Production Engineering, and Pedagogy. She has a PhD in Production Engineering from UNIMEP and a Master's degree in Production Engineering from UFSC - Federal University of Santa Catarina. Currently, she coordinates the undergraduate program in Railway and Metro Engineering. She has in-depth knowledge of Data Science and Analytics, and Digital Business (Business Intelligence). She is a professor at the Federal Institute of Education, Science and Technology of Southeast Minas Gerais - Juiz de Fora, MG. Since 2002, she has been an ad hoc evaluator of undergraduate courses (INEP/MEC).

Domingos Sávio da Cunha Garcia, Universidade Estadual de Campinas - UNICAMP, Campinas, SP

He holds a PhD in Applied Economics, specializing in Economic History, from the State University of Campinas (2005) and is an adjunct professor (C10) at the State University of Mato Grosso - UNEMAT, based in the History Department at the Cáceres campus since 1995. He has experience in the field of History, with an emphasis on Economic History and Political History of Brazil in the 19th century, working mainly on the following themes: political history of Brazil's western border during the long 19th century; geopolitics and international relations: Brazil and the River Plate in the 19th century; economy and society in France.

Leonardo Amorim de Araújo, Universidade Federal do Rio de Janeiro - UFRJ - Rio de Janeiro, RJ

He holds a degree in Civil Engineering from the Federal University of Juiz de Fora (1978), a master's degree in Transportation Engineering from Washington University in Saint Louis (1986), a doctorate in Transportation Engineering from the Federal University of Rio de Janeiro (2003), elementary school education from Ginásio Pio X (1968), and high school education from Instituto Metodista Granbery (1971). He is currently a tenured professor at the Federal Institute of Southeast Minas Gerais.

References

AGRAWAL, Ajay. The economics of artificial intelligence. McKinsey Analytics, abril de 2018. Disponível em: https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/the-economics-of-artificialintelligence. Acesso em 28/11/2025.

ALVES, P. C. P.; PRADO S. C. C. 2022. Estudo Comparativo entre algoritmos de machine learning aplicados à previsão de series temporal do mercado financeiro. In: IX Congresso de trabalhos de graduação faculdade de tecnologia e Mococa, 2022, São Paulo, SP, Brasil. Vol.9 N.1 A.2022.

BREIMAN, L., FRIEDMAN, J., STONE, C. J.; OLSHEN, R. A. (1984). Classification and Regression Trees. CRC Press.

GOODFELLO W, I., BENGIO, Y.; COURVILLE, A. (2016). Deep Learning. MIT Press. Disponível em: https://www.deeplearningbook.org/. Acesso em 28/11/2025.

HAN, J., KAMBER, M.; PEI, J. (2011). Data Mining: Concepts and Techniques. Elsevier.

HYNDMAN, R. J.; ATHANASOPOULOS, G. (2018). Forecasting: principles and practice. OTexts. Disponível em: https://otexts.com/fpp3/. Acesso em 28/11/2025.

Instituto de Segurança Pública. (2023). Relatório Anual de Estatísticas de Segurança Pública no Rio de Janeiro. Acesso em 2023, disponível em: www.isp.rj.gov.br. Acesso em 28/11/2025.

MEIJER, A.; WESSELS, M. Predictive policing: review of benefits and drawbacks. International Journal of Public Administration, v. 42, n. 12, p. 1031-1039, 2019. Disponível em: https://www.tandfonline.com/doi/full/10.1080/01900692.2019.1575664. Acesso em 28/11/2025.

PERRY, W. et al. Predictive policing: the role of crime forecasting in law enforcement operations. [s.l.]: Rand Corporation, 2013. Disponível em: https://www.rand.org/pubs/research_reports/RR233.html. Acesso em 28/11/2025.

PROVENZA, M. M.; COSTA, J. F. da S.; SILVA, L. D. C. Análise e previsão de séries temporais do homicídio doloso no Rio de Janeiro. Revista Cadernos de Estudos Sociais e Políticos, v. 4, n. 7, p. 63-85, 2015. Disponível em: https://www.e-publicacoes.uerj.br/index.php/CESP/article/view/18992/13810. Acesso em 28/11/2025.

ROEDER, O. K. et al. What caused the crime decline? New York: Brennan Center for Justice, 2015. p. 142. Disponível em: https://www.brennancenter.org/sites/default/files/publications/What_Caused_The_Crime_Decline.pdf. Acesso em 28/11/2025.

SILVA, L. J. S. et al. CrimeVis: an interactive visualization system for analyzing crime data in the State of Rio de Janeiro. In: INTERNATIONAL CONFERENCE ON ENTERPRISE INFORMATION SYSTEMS, 19., 2017, Porto, Portugal. Proceedings... Porto: Iceis, 2017. p. 193-200. Disponível em: http://www.scitepress.org/DigitalLibrary/Link.aspx?doi=10.5220/0006258701930200. Acesso em 28/11/2025.

TASHMAN, L. J. Out-of-sample tests of forecasting accuracy: analysis and review. International Journal of Forecasting, v. 16, n. 4, p. 437-450, 2000. Disponível em: https://linkinghub.elsevier.com/retrieve/pii/S0169207000000650. Acesso em 28/11/2025.