Análise do fluxo acadêmicoa dos alunos de engenharia da UFC: uma aplicação na cadeia de Markov

Abstract

The expansion of Brazilian higher education has brought new challenges to universities, es- pecially in terms of the permanence of students in undergraduate programs. The evasion and

retention of students is even more serious in public education, as it results in a waste of the union’s resources invested in the training of qualified professionals. Thus, the research and the utilization of tools that help understand student behavior in relation to this phenomenon is of great value. Locally, research has already been done in this field, such as evaluations of the effectiveness of teaching in one of the engineering courses at the UFC and the analysis of the motivations that influence the dropout of students through targeted questionnaires. This work aims to contribute to the study of this issue, through the analysis and modeling of the student flow of the Engineering courses of the Centro de Tecnologia (CT) of the Federal University of Ceará (UFC), using absorbing Markov chains. For this, a model was created with six transient states, which represent the non-enrollment and the quintiles of course completion, as well as three final states that contemplate graduation, course change and evasion. This method allows estimating some metrics such as the average time a student takes in each transient state, the time it takes a student to graduate, the probabilities of the student progressing or remaining in the same quintile, and the probabilities of the student reaching the final states. The analysis was performed on a dataset containing the enrollment information of students from semester 2015.1 to 2022.2, allowing to analyze the student behavior of students entering the same admissions exam and active students at the university in specific semesters. The results obtained show an average probability around 45% of an incoming engineering student to graduate, in addition to the distinct probabilities of student dropout at different stages of the course for the engineering majors of the UFC. Finally, estimates are presented that show a drop in the expected number of students graduating in the SISU 2016 and 2017 classes when compared to the results obtained in the 2015 class. The results obtained are of great value, as they provide important information to support the decision-making process of managers in the development of actions to improve the engineering courses.