Mineração de dados para análise de desempenho de alunos do Ensino Fundamental

• Autor(a) principal: Paula, Alexandre Abreu de
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Manancial - Repositório Digital da UFSM
• dARK ID: ark:/26339/0013000011bdg
• Texto Completo: http://repositorio.ufsm.br/handle/1/29311
• topic: Avaliação da Educação Básica no Brasil; Mineração de dados educacionais; Descoberta de conhecimento em bases de dados; Evaluation of Basic Education in Brazil; Educational data mining; Discovery of knowledge in databases; CNPQ::CIENCIAS HUMANAS::EDUCACAO

Nowadays, the challenge for the field of social policies is to improve the quality of basic education, and the Basic Education Development Index (Ideb) is the measure used to measure the performance of the Brazilian educational system. To face this challenge, Information and Communication Technologies (ICT) have proven to be a useful platform to support the learning process, generating large amounts of data. In state schools in Rio Grande do Sul, the Education Department's (ISE) computerization system is a tool that stores student information and generates data that can be used to discover new knowledge. The objective of this research is to generate a prediction model that allows identifying students with the potential to fail, so that the school can outline more focused actions for these students, enabling managers and teachers to apply solutions to improve student performance. The development of the work took place in three stages, with a deductive, qualitative and quantitative research method, which included a bibliographical research on the guiding theme and had as research universe 6th and 9th grade students from 2016 to 2019. At this stage, a study was carried out on the functioning of IDEB, the use of Data Mining techniques in Education (MDE) and the process of discovering knowledge in databases (KDD). In the second stage, stages of the KDD process were explored, applying data mining (DM) techniques and algorithms to identify attributes related to the low performance of 6th and 9th grade students. In the third stage, the data were evaluated and a web system was built to predict students. The results indicated that, in mining the 6th grade students' experiment, the greatest difficulty was found in the mathematics discipline, which is in line with the results obtained in IDEB. In the experiment with 9th grade students, the subject with the greatest difficulty was science. In both experiments, the pattern found by the algorithm as the root node was the 2nd quarter grade in math for the 6th grade and science for the 9th grade, allowing managers to take the necessary action to help students who might fail.