Discovery of the predictive success factors of students in the Logistics Technology Course, in ENADE exams, through Educational Data Mining

Abstract

The objective of this study was to discover, through Educational Data Mining, which factors were most associated with the best performances obtained by students of the Technology in Logistics course in the ENADE exams, of the 2018 edition. The data collected on the INEP website were treated and formatted in order to remove whites or nulls. The Educational Data Mining phase included the execution of the Decision Tree, Random Forest, Gradient Boosted Tree and Naive Bayes algorithms. After all tests were performed, the algorithm that showed the best performance was Naive Bayes with Accuracy = 98.21%, Kappa Index = 0.964, Recall = 83.32% and Precision = 82.40%. The results indicated that the factors related to the Number of hours for study, the level of education of the country, whether the Educational Institution provides adequate materials and equipment for classes, whether teachers use IT resources and whether the Course proposes Level Updated Knowledge, were more associated with better performance in ENADE assessments. The discovery of these factors can contribute to the development of action plans, by education professionals, that can propose improvements in the educational environment.