Teaching R with little time commitment


#1

I am a postdoc and I don't have a lot of time to teach (100% research), but I really want to help the grad students in our department learn R. So this semester, I tried out a teaching approach that required very little of my time and I was very happy with it!

I wrote about some of the details in a blog here, but thought I would also leave a post on RStudio Community where the idea would be easier to find.


#2

I applaud your approach for a volunteer effort at helping to pull up your successors. I think your approach on throwing them into the pool and being ready to throw a life ring after them was well considered given the constraints all around.

It's hard to think of a field (classical Greek literature?) where grad students aren't going to have to cope with ingesting, transforming and presenting data in all its forms even if they don't have to do more statistical manipulation on it than cross tabs. That's an essential skill set. Being able to do it programmatically, rather than through point-and-click will, as you save "change your life."

I've taken three edX courses this year that you could offer as a supplement. They're all free unless you want a gold star for good attendance and passing the tests. BerkeleyX is an undergraduate survey (in a particularly ugly domain-specific dialect of Python) of techniques of data science. MITx is a big-picture survey of the field of data science for those headed toward industrial-scale application, and HarvardX has a survey source of R, in nine courses, from basics through machine learning, with an text in process (https://goo.gl/UiwiuF) that provides a good guide. The lectures are illuminating. The tests are atrocious -- ambiguous questions, answer-bots untested against edge cases, etc., so I can't recommend paying for a certificate, but it will cover the waterfront for any grad student with a quant agenda to be covered.

A word about functions in R.

f(x) = y == y <- f(x)

Caution any of your future students with programming experience this R is simply advanced algebra, functions with arguments. Most people who walk away from R have tried and failed to map it to C and its descendants as an imperative/procedural language.


#3

Thanks for the kind words! These are great recommendations and insights!

I am moving towards also helping students get through online courses by offering credit and holding a grade over their head if they don't complete goals that they set out in the beginning of the semester. Hopefully this will work out with DataCamp. I have quite a few people ask me if they can take the one-credit R for Data Science course "online" even though we only meet once a week and the book is free online.

It's nice that there are so many resources out there now, but the people that I am targeting already have a lot on their plate and as much as they want to learn R, pressures of grades from other classes and deadlines from their advisors will win out over self-learning every time. My goal is to put their self-learning goals at a priority equal to other graded courses.