Fatores e evidências sobre o Exame Nacional do Ensino Médio (Enem): uma abordagem exploratória e experimental com mineração de dados

• Ano de defesa: 2021
• Autor(a) principal: Franco, Jacinto José
• Orientador(a): Brancher, Jacques Duílio
• Banca de defesa: Brancher, Jacques Duílio, Ferreira, Deller James, Góis, Lourival Aparecido
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Goiás
• Programa de Pós-Graduação: Programa de Pós-graduação em Ciência da Computação (INF)
• Departamento: Instituto de Informática - INF (RG)
• País: Brasil
• Palavras-chave em Português: Enem; Mineração de dados educacionais; Experimentos; Algoritmos de seleção
• Palavras-chave em Inglês: Enem; Educational data mining; Experiments; Feature selection
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://repositorio.bc.ufg.br/tede/handle/tede/11248

Enem is a test required by most Brazilian universities in order to select the best candidates to fill the vacancies available in undergraduate courses. Because it has been widely applied for 22 years, this assessment is an important thermometer of Brazilian education, thus allowing the identification of the factors that most contribute to student performance. The questionnaire applied at the time of Enem registration changes considerably over the years, but not everything that is collected is important to say whether the candidates' performance is high or low. Therefore, this work, based on this gap identified in the literature, we intend to list the most important factors for detecting high and low performance students. To achieve this goal, an experimental and exploratory approach is used regarding the use of attribute selection algorithms, validated with classifiers and replicated. The results indicate that from a limited set of 10 factors, it is possible to classify student performance with an average accuracy above 80\%. This result is interesting for researchers who aim to study Brazilian students, as up to 327 characteristics were collected in Enem in a single year. It is concluded that the combination of attribute selectors and classifiers allow the analysis of educational data to be facilitated, as it enables to focus on a restricted set of more significant factors. The main contributions of this work are constituted by the systematized experimental scheme, the experiments and patterns identified in the data, which subsidizes the realization of reflections on the student public. The main elements of this dissertation were condensed in an article, submitted and accepted for publication, the rest is in the process of being edited for publication in events in the area.