Better learning of data science in a biology curriculum by using R, RStudio, learnr & Github Classroom
Engels Guyliann & Grosjean philippe 2019-02-28
After switching to modern tools for teaching data science (http://biodatascience-course.sciviews.org) with R, RStudio, learnr & Github ClassRoom, a higher participation rate and better overall results were observed in comparison to the old, traditional biostatistics course. By using richer, interactive learning material, we were able to increase interest, participation and learning of undergraduate students enrolled in a biology curriculum at UMONS, Belgium. The content of the course increased by 40% without increasing the number of in-class hours (75h).
Among tools used, a fully configured virtual machine (svbox) with preinstalled R, RStudio, Python, Jupyter, Spyder and 1346 R packages is used both in class and at home. An e-book (bookdown) is used to collect pedagogical material in a centralized place. Twenty interactive tutorials (learnr) are proposed and individual learning progression is recorded (mongodb). Individual or group assessments are provided in Github Classroom. Data wrangling, plots and reproducible workflows are emphasized using tidyverse, and R Markdown.
85% of learnr exercises were finished, and a total of 187 Github repositories were created (37 students). Overall this approach was very successful with 92% success at the exam. We never got such a high success rate so far, with students that are not chiefly motivated to use a computer or to crunch numbers.