Best teaching platforms. Vary by audience?


I have two separate but related projects, both involving teaching data science principles with R. The first is more of a "typical" teaching environment, targeted at graduate students in the social sciences (mostly education, psychology, and business). The second is more experimental. I'm going to be working with middle school students in a math elective class.

I'm basically looking for advice on the best platform to use in each setting (maybe the same for both). RStudio Server? RStudio Cloud? RStudio Connect? Specifically with the middle school students, I don't want them to have to mess with installing anything, and I want the environment to be the same for everybody. I have no experience with any of the above, so I'm not sure on the similarities/differences, and which might be best suited for each task.


I've gotten in and poked around with RStudio Cloud a bit, and it looks like it will accomplish what you're looking for! I'm planning on using it myself for a summer workshop with high school students because of it's ease of use and accessibility, plus there's very little overhead (beyond creating the course and materials) on my part as an instructor.


Great, thanks! I will start there then.


I’ve been using Cloud in my college level intro DS course, and it’s great. I think it would work pretty well for the middle school case. And for the other one, if one of the learning goals isn’t software installation it should still work well. There are (pretty high) limits on how big datasets can be so depending on your use case you might check those out first. (I’ve not hit the limit.)


Thanks Mine, that’s really helpful. I think I’ll try it out and will report back to let you all know how it goes!


I did something similar re: grad student use case and I put out instructions for local RStudio installs. Then I built the course materials and made them pull it from my repository on github. However, I had an ulterior motivation as I wanted them to see the power of version control and leave with the software on their machine so they could apply it directly to their research.


Same here. RStudio Cloud doesn't seem to have very robust or easy to use support for Git yet. I repurposed Jenny Bryan's Happy Git with R instructions to setup R and Git. Students quickly adapt to the format.

Though I have always had pains with students installing certain packages for R, specifically setting up Spark with sparklyr. I'll have to see if RStudio Cloud makes that easier. Or I guess I could ask IT to setup an RStudio Server just for our course.


Thanks both, Medewitt makes a good point about not being "ready to go" after the course if the proper software is not installed. I think with grad students that's a sizable drawback to using something like RStudio Cloud. Of course, they could continue to use the cloud for their applied work. Maybe start with something like the cloud and help coach students on installs throughout the course and allow them to switch to their local whenever they feel comfortable? R Studio not having tight integration with git is a bit scary though. I'd like to run my graduate class through git starting this fall.


To clarify, RStudio (locally installed) has tons of support for Git. Once students install and setup Git through the terminal (to cache their login credentials), most of my students never touch the shell. Everything they need to do (clone, stage, commit, push, pull) can be done within RStudio's interface.

The drawback I saw for RStudio Cloud is that credentials are not stored and I wasn't sure how Git would work if students had to copy a project for each assignment. I create one repo per assignment and students fork from that to compete the work, then submit the final product as a pull request. I just haven't figured out how that would work from RStudio Cloud