Student dropout in higher education: the case of UFRN.
Dropout. Higher Education. Machine Learning.
The movement to expand access to higher education is being accompanied by a concern with the permanence of these students and with the obtaining of their undergraduate degrees, considering that dropout represents prejudice in the most varied aspects. In this perspective, the objective of this paper was to analyze the dropout of higher education students and to develop a predictive model of the risk of dropout of UFRN undergraduate students. To achieve this, an explanatory research was conducted with a quantitative approach, based on data provided by the “Superintendência de Tecnologia da Informação” of this university. Supervised Machine Learning Algorithms were used to develop the model. As a result, a tree-based prediction model was developed that explains dropout rates in relation to the UFRN scenario by an average of 77%. It was also observed that the number of failures was the variable with the greatest effect on evasion, according to the selected model.