Aplicação de técnicas de projeção multidimensional para análise do desempenho acadêmico dos alunos de graduação da FACOM-UFU

Abstract

In recent years, undergraduate courses in Information Technology (IT) have shown high failure rates, resulting in a decrease in the number of graduates in their respective cohorts. Analyzing and understanding the causes of these failures are essential not only for students undergoing the graduation process but also for educators and educational administrators. Information Visualization techniques stand out for their ability to reveal patterns and trends in data, aiding in the understanding of observed phenomena. These tools hold significant potential for application in educational environments. However, most existing studies rely on data related to students’ lives outside academic institutions, particularly socioeconomic data. This study employs information visualization techniques and graphical representation to investigate the most suitable representations of student data and analyze observed behavior patterns. The techniques were applied to anonymized data of undergraduates in Information Systems and Computer Science courses, collected by the Academic Control Division (DICOA) of the Administration and Academic Control Directorate (DIRAC) at the Federal University of Uberlândia (UFU), along with synthetic data created for testing specific patterns. By applying these visualization techniques, it is possible to observe and compare the quality and correlation between complete datasets and dimensionality reduction techniques, thereby facilitating analysis and enabling an accurate assessment of students’ performance in each administered course