PREVENDO O DESEMPENHO NO ENADE: UMA APLICAÇÃO DE ALGORITMOS DE APRENDIZAGEM DE MÁQUINA

Abstract

The objective of this study was to predict the performance of new students in the business administration course at Centro Universitario Cidade Verde (UNIFCV), based on the Student Questionnaire and data from the National Student Performance Exam (ENADE) 2018. used a database formed by socioeconomic and academic information submitted to Logistic Regression, Randon Forest, Decision Trees and Naive Bayes classifiers. The results showed advantages in the use of Logistic Regression in the analyzed data, presenting an accuracy of 63.04% and statistical significance. In addition, different individual characteristics presented, according to the Logistic Regression algorithm, different probabilities of the student achieving above average performance in ENADE.