This repo contains slides and demo materials for the talk
titled "Using R, RStudio, and Docker for introductory
statistics" at useR 2015
in Aalborg, Denmark.
If you want to find out more details about the Docker setup,
shoot me an email.
TL;DR
Why R?
Why / why not R?
- Why? Unlike many software designed specifically for
intro stat courses, R is
- free & open source,
- powerful & flexible, and
- very importantly relevant beyond the
intro stat classroom
- Why not R?
- teaching programming in addition to stats concepts
is challenging:
- option 1: don't do any hands on data analysis ->
terrible idea / disservice to students + faculty
- option 2: use a grad-and-drop type tool ->
end up with drag-and-drop instruction hell
- moral: there's a learning curve for any software,
if you're going to have your students climb up
some hill, make it worthwhile beyond just that
one class
- command line is intimidating: RStudio not so intimidating
How R?
- Technical:
- Short story: use RStudio, in the browser, avoid local
installation for seamless entry and minimized frustration
(by students + faculty)
- Each RStudio session lives in a Docker container, containers
persist for a semester (or more if you like)
- See architecture slide in presentation for implementation
details
- Pedagogical:
- Reproducibility is the goal, and the tool for teaching R
- Literate programming in R with RMarkdown
- Scaffolding: Start with very hand-holdy templates, slowly
take away the support, with end of semester goal of fully
reproducible data analysis projects
Resources:
- GitHub repo for Docker implementation at duke - updates to the repo forthcoming
- OpenIntro for R labs
- My github repo for
specific course details + materials for Sta 101 at Duke. Course
repos usually start with
sta101_
or sta104_
, latter is the
online version of the course. Each semester is a new repo,
there's lots of duplication between semesters.