| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.14 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
O enorme aumento na quantidade de dados educacionais gerados e armazenados possibilita obter informações de extrema importância sobre o processo de ensino e aprendizagem, nomeadamente informações dos estudantes que possam estar em risco de abandono ou que necessitam de atividades específicas para aumentar seu sucesso.
Neste sentido, o presente estudo busca identificar o perfil dos estudantes de licenciatura em engenharia da Escola Superior de Tecnologia e Gestão do Instituto Politécnico de Bragança (IPB) que abandonaram seus cursos e, a partir disso, propor um modelo de classificação do risco de abandono escolar baseado na árvore de decisão C5.0.
Para a execução deste estudo, foram utilizados dois conjuntos de dados obtidos a partir da base de dados Oracle do IPB, um conjunto para os estudantes diplomados e outro para os estudantes que abandonaram seus cursos (dropouts), ambos para o período de 2007 a 2015.
Manipular e analisar grandes volumes de dados costuma ser um processo lento e complexo e, por esse motivo, foram adotados métodos e técnicas de Academic Analytics, em busca de compreender os dados utilizados neste estudo.
Os principais resultados permitem concluir que a maior parte dos dropouts ocorre antes mesmo de os estudantes estarem registrados por um ano em seus cursos. Em muitos casos, os estudantes que abandonam possuem um rendimento semelhante ao dos diplomados, o que pode indicar a existência de fatores externos ao IPB que favorecem ao abandono. O número total de dropouts em cada curso é no mínimo 30% dentro do intervalo de 9 anos. 25% dos estudantes abandonaram seus cursos sem ter cursado ao menos uma disciplina.
A criação do modelo baseado em árvore de decisão tem como principais objetivos a geração de instruções padronizadas, facilidade de interpretação e permitir a adição de vários cenários possíveis, contribuindo com diversas vantagens ao processo de tomada de decisão.
The enormous increase in the amount of educational data generated and stored enables us to obtain extremely important information about the teaching and learning process, namely information from students who may be at risk of dropout or who need specific activities to increase their success. In this sense, the present study aims to identify the profile of the engineering students of the Polytechnic Institute of Bragança (IPB) who dropped out and, from this, propose a risk classification model of school drop-out based on the decision tree C5.0. For the execution of this study, we used two data sets obtained from the IPB Oracle database, one for the graduate students and another for the dropout students, both for the period from 2007 to 2015. Manipulating and analyzing large volumes of data is often a slow and complex process and, for this reason, Academic Analytics methods and techniques have been adopted in order to understand the data used in this study. The main results allow us to conclude that most dropouts occur before students are enrolled for one year in their courses. In many cases, students who drop out have a similar performance to the graduates, which may indicate the existence of factors external to the IPB that favor dropout. The total number of dropouts in each course is at least 30% over the 9 years. 25% of students dropped out without attend at least one subject. The creation of the model based on decision tree has as main objectives the generation of standardized instructions, easy interpretation and allow the addition of several possible outcomes, contributing with several advantages to the decision-making process.
The enormous increase in the amount of educational data generated and stored enables us to obtain extremely important information about the teaching and learning process, namely information from students who may be at risk of dropout or who need specific activities to increase their success. In this sense, the present study aims to identify the profile of the engineering students of the Polytechnic Institute of Bragança (IPB) who dropped out and, from this, propose a risk classification model of school drop-out based on the decision tree C5.0. For the execution of this study, we used two data sets obtained from the IPB Oracle database, one for the graduate students and another for the dropout students, both for the period from 2007 to 2015. Manipulating and analyzing large volumes of data is often a slow and complex process and, for this reason, Academic Analytics methods and techniques have been adopted in order to understand the data used in this study. The main results allow us to conclude that most dropouts occur before students are enrolled for one year in their courses. In many cases, students who drop out have a similar performance to the graduates, which may indicate the existence of factors external to the IPB that favor dropout. The total number of dropouts in each course is at least 30% over the 9 years. 25% of students dropped out without attend at least one subject. The creation of the model based on decision tree has as main objectives the generation of standardized instructions, easy interpretation and allow the addition of several possible outcomes, contributing with several advantages to the decision-making process.
Description
Dupla diplomação com a Universidade Tecnológica Federal do Paraná
Keywords
Ensino superior Abandono escolar Engenharia Mineração de dados Academic analytics