Rede Bayesiana para Previsão de Evasão Escolar
Tendo em vista o problema de evasão
escolar, problema este que impacta instituições
de ensino do mundo inteiro, este artigo apresenta
a aplicação de Redes Bayesianas, com o
intuito de predizer os percentuais de chance de
evasão dos alunos, com o objetivo de auxiliar
os gestores educacionais na prevenção destes
tipos de situações. A predição é realizada com
base nas características dos alunos, coletadas
do sistema utilizado pelo SENAI de Tubarão/SC.
É possível ainda manipular tais características,
por parte do gestor, a fim de simular cenários com
o objetivo de minimizar as chances de o aluno
evadir. Através da validação dos resultados foi
obtido 85,6% de taxa de acerto, o que indica um
bom desempenho da rede bayesiana modelada
para o sistema desenvolvido.
PALAVRAS-CHAVE: Evasão escolar, modelo
preditivo, classificador bayesiano
ABSTRACT: Given the problem of school
evasion, a problem that affects educational
institutions around the world, this paper presents
the application of Bayesian networks to predict the
percentage chance of student evasion, in order
to assist educational managers in preventing
these types of situations. The prediction model
was built based on the characteristics of the
students, collected from the data base system
used by SENAI (Tubarão/SC). It is possible
to manipulate these characteristics, by the
manager, in order to simulate scenarios to
minimize the chances of the student evasion.
Through the validation of the results, we have
obtained 85.6% of accuracy, which indicates
a good performance of the Bayesian network
modeled for the system developed.
Rede Bayesiana para Previsão de Evasão Escolar
-
DOI: 10.22533/at.ed.7461922043
-
Palavras-chave: Evasão escolar, modelo preditivo, classificador bayesiano
-
Keywords: School evasion, predictive model, bayesian classifier
-
Abstract:
Given the problem of school
evasion, a problem that affects educational
institutions around the world, this paper presents
the application of Bayesian networks to predict the
percentage chance of student evasion, in order
to assist educational managers in preventing
these types of situations. The prediction model
was built based on the characteristics of the
students, collected from the data base system
used by SENAI (Tubarão/SC). It is possible
to manipulate these characteristics, by the
manager, in order to simulate scenarios to
minimize the chances of the student evasion.
Through the validation of the results, we have
obtained 85.6% of accuracy, which indicates
a good performance of the Bayesian network
modeled for the system developed.
-
Número de páginas: 15
- Willian Silvano Maria
- João Lucas Damiani
- Max Roberto Pereira
- Max Pereira