Mineração de dados educacionais para predição de desempenho

• Autor(a) principal: Silva, Daniele Kretli
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da ESPM
• Texto Completo: http://tede2.espm.br/handle/tede/633
• topic: mineração de dados educacionais; estilos de aprendizagem; traços de personalidade; desempenho acadêmico; CIENCIAS SOCIAIS APLICADAS::ADMINISTRACAO

The purpose of this work is to investigate how different learning styles and personality traits could be considered as predictors of academic performance enabling the development of predictive performance models. It is proposed that with the input of a new student's data, it will be possible to identify the probability of high or low performance in the knowledge areas of the courses available. Historical information from students of a private higher education institution were collected and Linear Regression statistical techniques were used to analyze the relationship between academic performance and demographic variables, learning styles and personality traits. The model was built using the Naive Bayes classification algorithm, which calculates the probability of a student having high or low performance in certain knowledge areas based on individual characteristics. The methodology chosen to implement the knowledge discovery process was CRISPDM (CRoss-Industry Standard Process for Data Mining), including the steps for planning, organizing, implementing and documenting a data mining project. As a result, evidence shows that female students performed better than male students in all knowledge areas explored. In addition, both in terms of learning styles and personality traits, we have evidence that the role of these characteristics can impact performance across the knowledge areas of the courses. Therefore, it was possible to create an academic performance prediction model based on the individual characteristics of the student.