Exécution de notre outil de synthèse sur un nouvel article

Résumé : In recent years, online courses have become very popular due to the restrictions on the movement of people resulting from the COVID pandemics. The expansion of online learning platforms has renewed interest in developing techniques for delving into student behavior and improving their learning experience. This paper proposes a novel analysis of data obtained from Moodle’s Activity Log using Learning Analytics techniques and different clustering algorithms. The main goal is to understand the students’ task-related behavior, to generate a comprehensible model, and to comprehend how it is related to their academic performance. The students were grouped using a single feature calculated for each task the students must complete in the course. The results show that this single feature is enough to model the students’ behavior and their academic performance. Different clustering algorithms and number of clusters were used, and K-Means algorithm with three clusters achieved the best result. These three groups are comprehensible and explain the students’ behavior according to their activity in the task’s submission. The clusters also have a direct relation to the students’ final grade in the course, being one of them composed of students at risk of failure.