Use este identificador para citar ou linkar para este item: http://104.156.251.59:8080/jspui/handle/prefix/880
Título: Tipificação de ocorrências policiais utilizando machine learning
Autor(es): Reis, Marcelo Indio dos
lattes.cnpq.br
Reis, Marcelo Indio dos
lattes.cnpq.br
Melo, Osvaldo Requião
lattes.cnpq.br
Aquino, Pamela Arielle Brito de
lattes.cnpq.br
Palavras-chave: Aprendizado de máquina
Ocorrência policial
Classificação
Machine learning
Police report
Classification
Data do documento: 18-Jun-2019
Editor: Universidade Católica do Salvador
Resumo: Public safety is one of the main pillars for society that directly influence the citizens quality of life. Lately this area has been receiving a great focus on the investment issue, as Ballesteros (2014) says his article, much of this value has been destined to process automation. In this context, technology enters to support various routine activities, making them more efficient through better management of their resources. With that, it was suggested in this assignment the use of machine learning algorithms, to automatically typify police occurrences. It was used a base of data with police records and from it some samples were taken for validation. For this tests were selected the algorithms C4.5, CART, KNN, SVM, Rede Neural, Ripper and Random Forest. The results gotten, indicated that the C4.5 and the Ripper had the best accuracy, becoming until 99% in some tests. In a test, the application of the coefficient of correlation of Matthews and of F1 score, reached 0.89 and 0.94 in that order.
URI: http://104.156.251.59:8080/jspui/handle/prefix/880
Aparece nas coleções:Engenharia de Software

Arquivos associados a este item:
Arquivo Descrição TamanhoFormato 
TCCMATHEUSEJUAN.pdf867.77 kBAdobe PDFVisualizar/Abrir
Juan_Matheus (anexos).zip3 kBZIP archiveVisualizar/Abrir


Os itens no repositório estão protegidos por copyright, com todos os direitos reservados, salvo quando é indicado o contrário.