Avaliação Antecipada do Risco de Desistência em Disciplinas de Programação da ECT/UFRN: uma abordagem de aprendizado de máquina supervisionado e não supervisionado

Abstract

This study aims to apply machine learning techniques to predict the academic performance of students in programming courses at the School of Science and Technology at the Federal University of Rio Grande do Norte (ECT/UFRN). Using data collected from the LOP platform, supervised and unsupervised approaches were employed to analyze student behavior. Azure Machine Learning was used to identify the best classification model, while Google Collaboratory was utilized to uncover natural patterns in the data using the K-Means algorithm. The analyses provide valuable insights that can guide pedagogical interventions to improve students' academic performance. Future research may explore more advanced techniques and the integration of additional data to enhance predictions.