I worked for a few years with academics who insisted on using SAS. It was a new
language for me but as I became more experienced, I grew to really enjoy using
it. Since my transition to consultancy – sans available SAS license (FYI: it is
expensive!) – I’ve set myself out to learn R.
After just one project, here are a few rambling thoughts I’ve had on R and SAS:
- RStudio helped me immensely in my transition into R. The setup is quite
reminiscent of the SAS layout and being able to view my code, the console
(where the code is executed), my datasets, the output and more in one window
is a huge improvement over the basic R setup.
- Base R? Ok, yeah it’s good. But R packages are amazingly useful. The
difficulty: which one do you need? I’ve found it’s really taken the guidance
of experienced R users to find the most useful package for whatever task is
at hand. Sometimes googling will help; there will always be a Stack Overflow
thread to be found. But you might also find five different answers.
- Speaking of packages, the
dplyr
package that pipes a dataset through
various functions strongly reminds me of data and proc statements in SAS.
Before learning dplyr
, I found modifying data in R to be much more
laborious than in SAS. Now dplyr
is one of my favourite packages. - SAS has a lot of functions. Like a lot. And because they’re all written and
managed by the company, syntax across functions is usually quite consistent.
Given that R packages are user written, syntax can sometimes seem a little
wonky. However, while SAS syntax is reliably consistent, updates and new
functions are not regularly available; whereas that funny R package might
soon be improved and new ones are always appearing.
- Graphics produced with
ggplot2
(yes, another package) are far more
sophisticated than anything I managed to create in SAS. Bonus: it took me
about an hour to learn the basics of ggplot2
(incidentally, basic ggplot2
graphics are fabulous) whereas anything more than basic (and super boring)
graphics in SAS still seem somewhat opaque to me. - RMarkdown
, available to use with Rstudio, can be used to produce complete
documents, reports, etc., including figures and text, directly from your
code. How cool is that? From SAS, I managed to output highly customized
tables (requiring A LOT of coding) and decent graphics, but always needed to
insert text and finalize formatting post-export. Extra bonus: RMarkdown is
super, duper simple.
I could go on, but my ramblings appear to be mutating (dplyr
joke!) into a
pro-R list. Oops. I’ll finish up by saying that I feel as though I’ve just
scratched the surface of what I can do with R and I know I’ll have fun learning
more every day.