Temporal analysis of academic performance in higher education before, during and after COVID-19 confinement using artificial intelligence

This study provides the profiles of students and a regression prediction of marks considering data before, during, and after the COVID-19 pandemic. Using a field experiment of 396 students and more than 7400 instances, we have studied the differences in performance considering the temporal distribution of autonomous learning during courses from 2016/2017 to 2020/2021. After applying unsupervised learning, results show 3 profiles of students: continuous students, last-minute students, intense and last-minute students. We have found that the highest success ratio is related to students that work in a continuous basis. However, last minute working is not necessary linked to failure. After applying regression, results show that the mark of the students can be predicted successfully and that the most relevant values are the mean mark in self-evaluation obtained the month before the final exam, the mean mark two months before, the number of attempts two months before and the number of attempts the month before. Results are a little worse but still acceptable if the prediction wants to be made a month before the final exam. This regression is useful to prevent students' wrong learning strategies, and (more effective) to detect malpractices such as copying. We have done all these analysis taking into account the effect of the COVID-19 pandemic, including also a discussion about which factors will be extended in time and which ones are transitory and only due to the confinement.