Teaching data science to students in biology using R, RStudio and Learnr: Analysis of three years data

Abstract

We examine the impact of implementing active pedagogical methodologies in three successive data science courses for a biology curriculum at the University of Mons, Belgium. Blended learning and flipped classroom approaches were adopted, with an emphasis on project-based biological data analysis. Four successive types of exercises of increasing difficulties were proposed to the students. Tutorials written with the R package learnr were identified as a critical step to transition between theory and the application of the concepts. The cognitive workload needed to complete the learnr tutorials was measured for the three courses and it was only lower for the last course, suggesting students needed a long time to get used to their software environment (R, RStudio and git). Data relative to students' activity, collected primarily from the ongoing assessment, were also used to establish student profiles according to their learning strategies. Several suboptimal strategies were observed and discussed. Finally, the timing of students contributions, and the intensity of teacher-learner interactions related to these contributions were analyzed before, during and after the mandatory distance learning due to the COVID-19 lockdown. A lag phase was visible at the beginning of the first lockdown, but the students' work was not markedly affected during the second lockdown period which lasted much longer.