F6-COMP6-2 - Undergraduate Students' Effectiveness in an Institution With High Dropout Index

3. Research Full Paper
João Lucas Oliveira1 , Ana Paula Ambrósio1, Uyara Silva1, Jacques Brancher2, Jacinto Franco1
1 Universidade Federal de Goiás
2 Universidade Estadual de Londrina

Currently, the dissemination of open data in conjunction with EDM (Educational Data Mining), learning analytics, e-learning, enhanced technology learning, intelligent systems, intelligent tutors, data science, machine learning and online judge techniques made useful contributions to the field of education through knowledge generated from data analysis. Identifying factors that allow us to understand how students learn and their behavior helped managers and teaching professionals to identify the best teaching configurations inside and outside the classroom, which is still an open challenge with a lot of involvement from researchers. This study aims to compare the academic success of undergraduate students with other studies in the literature that also study academic success and retention, currently the university has been suffering from a high student retention rate, which motivated this research. The central idea is to know if there is any pattern among high-performance and low-performance students and if that pattern tends to change in different universities that also have their student analysis studies published. Two thousand four hundred and ninety-nine students (2499) were analyzed for more than 11 years. These students belong to one of the 20 best Brazilian universities and to three undergraduate courses in computing (Computer Science, Software Engineering and Information Systems), which allowed us to analyze heterogeneous cohorts. Statistical and data mining techniques were used to extract information that can validate the hypotheses of this study. Our goal is to find out what factors tend to contribute to students' retention, dropout, difficulties, engagement, motivations and academic success. To achieve this goal, we compared gender effectiveness and the course curriculum. The data showed that some factors, not previously analyzed by other studies, tend to influence student performance; the results can contribute to a comparison with other studies that also use data as a source of knowledge to parameterize their teaching contexts and find flaws in the process of building students' learning in the classroom.