No abstract

The present article is aimed at analyzing the effects of learning analytics on students' self-regulated learning in a flipped classroom. An experiment was conducted with 96 engineering students, enrolled in a subject offered in the Flipped Classroom model. The students were divided into two groups: an experimental group (N = 51) and a control group (N = 45). After each learning unit, while the control group did not have access to any learning analytics resources, students from the experimental group received a bulletin with feedback to support Self-Regulated Learning. The levels of student self-regulation were measured through questionnaires at the beginning and the end of the course. The analysis of the collected data revealed that the bulletin promoted significant effects in self-regulated learning in the experimental group, stimulating the self-reflection and colleague's support search for clarifying doubts. These results demonstrate that learning analytics can be used to promote self-regulated learning in flipped classrooms, helping students identify strategies that can increase their academic performance.

No abstract

The use of machine learning and data mining algorithms in educational contexts has evolved due to the large availability of data generated mainly in virtual learning environments. This study makes a comparative analysis of five classifiers in the task of predicting students with risk of dropping out in undergraduate courses by distance education. The results showed a small advantage for the use of Logistic Regression in the data analyzed, with success rates above 90% in the predictive model. Resumo.O uso de algoritmos de mineração de dados e de aprendizagem de máquina em contextos educacionais tem evoluído em razão da grande disponibilidade de dados geradas principalmente em ambientes virtuais de aprendizagem. Este estudo faz uma análise comparativa de cinco classificadores na tarefa de predição de alunos com risco de evasão em cursos de graduação por EAD. Os resultados apontaram uma pequena vantagem para o uso da Regressão Logística nos dados analisados, com taxas de sucesso acima de 90% no modelo preditivo.

No abstract