The number of resources for learning R has exploded in the past few years. Here are a few quick thoughts about how I would approach learning R in 2019 vs. 2015
I used Stata when I was an undergraduate economics student at VCU. When I graduated from college, my license expired and I had no money so I switched to R because it was free. It was one of the best decisions of my life.
A picture of me after graduation
There were fewer resources for learning R in January 2015 than January 2019. Knowing what I know now, I would approach learning R differently than I did when I graduated.
How I learned R
Like thousands of other people, I learned R through the Coursera Data Science Specialization. This nine course sequence is a rapid-fire survey of statistical, programming, and data science techniques taught by Roger Peng, Brian Caffo, and Jeff Leek.
I sought opportunities, mostly unsuccessfully, to use R at my various jobs after college. It didn’t improve my R skills but I had lots of opportunities to “pitch” R, which has benefitted me to this day. Instead, I wrote lots of terrible R code on my own time. In one project, I looked at the cost of alcohol at Virginia ABC stores. In another project, I plotted the continents of finishers at the UCI World Road Championships in Richmond, Virginia.
Finally, I accepted a job at the Urban Institute and moved to Washington, DC. Truly learning R wasn’t possible until I had a venue to tackle applied questions for hours each day. I didn’t know R until probably six months after joining the Urban Institute, which was two years after I started programming in R.
How I would learn R today
The data science specialization changed my life. In hindsight though, I would start with the tidyverse and narrow the scope of my studies. R for Data Science is too clear and too free for it to not be the first resource for anyone interested in learning R. I would read that book and do every exercise. Then I would read it again and probably do every exercise again. I would print off every cheat sheet and tape them to my wall. Then I would begin mastering one of the tools: data visualization, data reporting, or modeling. Finally, I would learn base R, which is still important, from a book like Norm Matloff’s The Art of R Programming.
My biggest challenge learning R was that I was isolated. For more than a year, I didn’t know any other serious R programmers. That’s like trying to learn Cantonese without knowing anyone who speaks Cantonese.
In hindsight, I should have worked harder to find other R programmers. This is easier today than ever before because of sites like Meetup and RStudio Community. Universities are also embracing R, so I would probably reach out to professors or administrators at my university.
To augment my search, I would start following the #rstats tag on Twitter. It is the most interesting, generous, and civil corner of Twitter and is a great way to get to know the community.
Finally, I would start an RSS feed for R blogs. Blogdown sparked an explosion of blogs in the R community, but there were plenty of valuable blogs in 2015 that would have helped my R programming. I would follow them all, start reading lots of posts, finish only a handful of posts, and bookmark the ones that solve my problems.