Análise sobre a evasão de alunos do curso de Estatística na FCT/Unesp
Carregando...
Data
2023-01-27
Autores
Orientador
Silvestre, Miriam Rodrigues
Coorientador
Pós-graduação
Curso de graduação
Estatística - FCT
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Tipo
Trabalho de conclusão de curso
Direito de acesso
Acesso aberto
Resumo
Resumo (português)
Um dos grandes problemas que aflige a educação há uma longa data até a atualidade é a evasão escolar, principalmente no ensino superior. Essa é uma realidade também no curso de graduação em Estatística na FCT/UNESP de Presidente Prudente - SP. Visando identificar os perfis de alunos mais propensos a evadirem do curso, foram utilizadas duas técnicas estatísticas neste trabalho. A regressão logística e a árvore de decisão são técnicas de classificação, ou seja, a partir delas conseguimos prever uma variável categórica, no caso do presente estudo, se o aluno irá ou não evadir do curso. Essas técnicas foram aplicadas em uma base de dados com variáveis sociodemográficas, como idade, sexo, tipo de escola cursada no ensino médio, etc., de alunos ingressantes entre os anos de 2008 e 2017 no curso de Estatística, com o objetivo de terem seus resultados comparados. As análises apresentadas foram realizadas no software R. Ao comparar as duas técnicas pela métrica F1-score, a árvore de decisão se ajustou melhor no treino, porém obteve-se maior poder de previsão na regressão logística.
Resumo (inglês)
One of the major problems that has afflicted education for a long time until today is school dropout, especially in higher level of education. This is also a reality in the undergraduate course in Statistics at FCT/UNESP in Presidente Prudente - SP. In order to identify the profiles of students most likely to dropout of the course, two statistical techniques were used in this work. Logistic regression and decision tree, which are classification techniques, that is, from them we can predict a category variable, in the case of this study, whether or not the student will dropout of the course. These techniques were applied to a database with sociodemographic variables, such as: age, gender, type of high school attended, etc., of students entering the Statistics course between 2008 and 2017, with the aim of having their results compared. The analyzes presented were performed in the programming language R. When comparing the two techniques by the F1-score metric, the decision tree fitted better in training, but greater predictive power was obtained in logistic regression.
Descrição
Idioma
Português