Aplicação de Mineração de Dados para Extração de Conhecimento de Crimes de Violência Doméstica: um Estudo de Caso na Cidade de Campo Grande (MS)

Abstract

The amount of data generated and stored has been increasing along with the computational power for storing it. For this data to become useful information for companies and people in decision-making positions, some data mining techniques may be applied. Through them, it is possible to find information, associations, and patterns regarding the analyzed data, which may serve both to extract knowledge and to make predictions. Public administration bodies responsible for public safety can benefit from data mining to enhance their actions in fighting and preventing crime, as can be seen in many places around the world. In the state of Mato Grosso do Sul, data on police occurrences are stored in the Integrated Operational Management System (SIGO). However, they have not yet been investigated for the extraction of crime knowledge or more effective decision-making. Thus, the objective of this project is the use of data mining techniques for knowledge extraction, considering the information stored in the SIGO system, more specifically, those regarding domestic violence registered in the of Campo Grande (MS) city. Furthermore, to improve the explanation and interpretation of the results for decision makers, symbolic data mining algorithms were used: association and classification rules. With them, it was possible to extract interesting and innovative knowledge from the data. For example, analyzing the insult fact, it can be observed that the victims are more often those with higher education. As for the classification rules, an accuracy of 84% was obtained, allowing the extraction of knowledge such as: victims aged higher or equal than 23 years register 1 to 3 police reports against the same author.