Modelo de gestão preditivo evasão em perspectiva para o processo decisório de instituições de ensino
Carregando...
Tipo de documento
Tese
Data de publicação
2025
Departamento
Pró-Reitoria de Pós-Graduação, Pesquisa e Extensão
Curso
Doutorado Profissional em Educação e Novas Tecnologias
Local de publicação
Curitiba/PR
Modalidade de acesso
Acesso aberto
Autor(es)
Silva, Moacir Gomes da
Orientador(es)
Alves, Alceli Ribeiro
Coorientador(es)
Membros da banca
Alves, Alceli Ribeiro
Bernardi, Luci dos Santos
Bragagnolo, Adriana
Lopes, Luís Fernando
Medeiros, Luciano Frontino
Santos, Rodrigo Otavio dos
Viana, Helena
Resumo
Esta tese, concebida em um programa que tem por área de concentração: Educação
e Novas Tecnologias, se propõe a desenvolver um Modelo de Gestão Preditivo
Evasão em Perspectiva, para contribuir no processo decisório de profissionais da
educação no enfrentamento da evasão escolar. Em particular, uma tese desenvolvida
no âmbito da Cátedra Unesco de Cidades que educam e transformam, Redes Unitwin,
trazendo para a discussão a importância e o papel das instituições de ensino superior
(como espaços formais de educação que são) na formação de professores e
profissionais da área de educação, de forma associada às tecnologias educacionais
que favorecem não apenas o acesso à educação à pessoas localizadas nas regiões
mais remotas ou periféricas do país, mas também instrumentos capazes de mensurar
os fatores de evasão, refletindo e perpassando sobre as estratégias de retenção, com
o intuito de assegurar o pleno exercício do direito à educação. O delineamento da
pesquisa contempla direcionamentos da Design Science, que por sua vez se utiliza
da Design Science Research como método que fundamenta e operacionaliza a
condução de pesquisas que têm como objetivo a construção de um artefato. O estudo
teve como objetivo geral desenvolver um modelo de gestão, com simulação
computacional, para predizer alunos em risco de evasão e contribuir para o processo
decisório das IES na mitigação deste fenômeno. Os objetivos específicos foram: 1)
mapear os modelos conceituais de evasão, tecnologia e gestão; 2) conceber modelo
de gestão preditivo evasão em perspectivas para o processo decisório; 3) analisar a
base de dados e os resultados da IES pesquisada com a aplicação do modelo de
gestão preditivo evasão em perspectiva. Os procedimentos metodológicos seguem os
aspectos da pesquisa quantitativa e qualitativa, de caráter exploratório, indutiva,
descritiva, aplicação de modelagem, fundamentação bibliográfica e revisão
sistemática da literatura inspirada e norteada pelos pressupostos da Design Science
Research (DSR). Os resultados são apresentados a partir de figuras, tabelas e telas
do simulador computacional. O produto obtido foi o desenvolvimento de simulador
computacional preditivo, de evasão discente, associado a uma modelagem inovadora
de nome Evasão em Perspectiva, que permite as instituições de educação superior,
através de mapa de desempenho do aluno, de forma antecipada, desde o momento
da matrícula do aluno na instituição, até sua formatura, em tempo real, compreender
o comportamento possível de sua evasão ou permanência escolar e assim definir suas
estratégias de gestão. Os principais stakeholders contemplados nesta tese:
pesquisadores, alunos, IES e seus gestores. Aplicou-se o modelo de gestão preditivo
em 27 variáveis, quantitativas e qualitativas, nas 4 perspectivas propostas, para
43.814 alunos da modalidade de educação à distância (EAD) do curso de licenciatura
em Pedagogia da IES pesquisada no período de 2019 a 2024, apresentando taxas de
evasão de 52,2% e as variáveis da perspectiva acadêmica e financeira apresentaramse como dominantes nos fatores de estímulo a evasão. O simulador obteve acurácia
de 85,4% com o algoritmo Naive Bayes e os recursos do Weka.
This thesis, conceived in a program that has as its area of concentration: Education and New Technologies, proposes to develop a Predictive Management Model for Dropout in Perspective, to contribute to the decision-making process of education professionals in tackling school dropout. In particular, a thesis developed within the scope of the UNESCO Chair of Cities that Educate and Transform, Unitwin Networks, bringing to the discussion the importance and role of higher education institutions (as formal educational spaces that they are) in the training of teachers and professionals in the area of education, in a way associated with educational technologies that favor not only access to education for people located in the most remote or peripheral regions of the country, but also instruments capable of measuring dropout factors, reflecting and going through retention strategies, with the aim of ensuring the full exercise of the right to education. The research design includes guidelines from Design Science, which in turn uses Design Science Research as a method that supports and operationalizes the conduction of research that aims to construct an artifact. The study had as its general objective to develop a management model, with computer simulation, to predict students at risk of dropping out and contribute to the decisionmaking process of HEIs in mitigating this phenomenon. The specific objectives were: 1) to map the conceptual models of dropout, technology and management; 2) to design a predictive management model for dropout in perspectives for the decision-making process; 3) to analyze the database and the results of the HEI studied with the application of the predictive management model for dropout in perspective. The methodological procedures follow the aspects of quantitative and qualitative research, of an exploratory, inductive, descriptive nature, application of modeling, bibliographic foundation and systematic review of the literature inspired and guided by the assumptions of Design Science Research (DSR). The results are presented using figures, tables and screens from the computer simulator. The product obtained was the development of a predictive computer simulator for student dropout, associated with an innovative model called Dropout in Perspective, which allows higher education institutions, through a map of student performance, in advance, from the moment the student enrolls in the institution until graduation, in real time, to understand the possible behavior of their dropout or permanence in school and thus define their management strategies. The main stakeholders covered in this thesis are: researchers, students, HEIs and their managers. The predictive management model was applied to 27 quantitative and qualitative variables, in the 4 proposed perspectives, for 43,814 students of the distance learning modality (EAD) of the Pedagogy degree course at the HEI researched in the period from 2019 to 2024, presenting dropout rates of 52.2% and the variables from the academic and financial perspective were presented as dominant in the factors stimulating dropout. The simulator obtained an accuracy of 85.4% with the Naive Bayes algorithm and Weka resources.
This thesis, conceived in a program that has as its area of concentration: Education and New Technologies, proposes to develop a Predictive Management Model for Dropout in Perspective, to contribute to the decision-making process of education professionals in tackling school dropout. In particular, a thesis developed within the scope of the UNESCO Chair of Cities that Educate and Transform, Unitwin Networks, bringing to the discussion the importance and role of higher education institutions (as formal educational spaces that they are) in the training of teachers and professionals in the area of education, in a way associated with educational technologies that favor not only access to education for people located in the most remote or peripheral regions of the country, but also instruments capable of measuring dropout factors, reflecting and going through retention strategies, with the aim of ensuring the full exercise of the right to education. The research design includes guidelines from Design Science, which in turn uses Design Science Research as a method that supports and operationalizes the conduction of research that aims to construct an artifact. The study had as its general objective to develop a management model, with computer simulation, to predict students at risk of dropping out and contribute to the decisionmaking process of HEIs in mitigating this phenomenon. The specific objectives were: 1) to map the conceptual models of dropout, technology and management; 2) to design a predictive management model for dropout in perspectives for the decision-making process; 3) to analyze the database and the results of the HEI studied with the application of the predictive management model for dropout in perspective. The methodological procedures follow the aspects of quantitative and qualitative research, of an exploratory, inductive, descriptive nature, application of modeling, bibliographic foundation and systematic review of the literature inspired and guided by the assumptions of Design Science Research (DSR). The results are presented using figures, tables and screens from the computer simulator. The product obtained was the development of a predictive computer simulator for student dropout, associated with an innovative model called Dropout in Perspective, which allows higher education institutions, through a map of student performance, in advance, from the moment the student enrolls in the institution until graduation, in real time, to understand the possible behavior of their dropout or permanence in school and thus define their management strategies. The main stakeholders covered in this thesis are: researchers, students, HEIs and their managers. The predictive management model was applied to 27 quantitative and qualitative variables, in the 4 proposed perspectives, for 43,814 students of the distance learning modality (EAD) of the Pedagogy degree course at the HEI researched in the period from 2019 to 2024, presenting dropout rates of 52.2% and the variables from the academic and financial perspective were presented as dominant in the factors stimulating dropout. The simulator obtained an accuracy of 85.4% with the Naive Bayes algorithm and Weka resources.
Palavras-chave
Evasão, Design Science, Gestores de IES, Modelos de Gestão
