I’ve used SAS statistics software for 30 years now. In a few years I’ll retire, and lose my workplace access to SAS (which I think costs a couple thousand dollars per year). But I want to be able to continue having the same or comparable capability. Should I learn R? How valuable would it be to learn R anyway, on the basis it might be better to use in these next few years?
I do fun things in SAS like this (in addition to my proprietary day job):
Using surveying data including radius, altitude, azimuth, and GPS coordinates in a regression model to create maps.
Using historical datasets from multiple sources in multiple formats to create timelines.
Creating diverse groups of people from demographic data using an iterative method.
Estimating dimensions of photographed objects by modeling pixel brightness as a function of X and Y, with iterative nonlinear modeling.
Working out the angular geometry to calibrate telescope pointing from a dataset of the declinations and right ascensions of bright stars in a regression model.
Decoding GPS data streams (like the NMEA sentences) or maybe even postprocessing pseudorange data.
I have used R, both professionally and for personal use. I quite like it, although you need to think a bit differently about your programs than if you were using Java or C.
I have not used SAS, so I can’t offer a perspective there.
R can be used for a wide variety of applications. I don’t specifically know about all of yours but I would be unsurprised to find out they can all be done.
Unless you have some reason not to, I think R would be a fine addition to your toolset.
I’m an occasional SAS user. Mostly I ask other people to do the SAS work these days, which is kind of a bummer. Anyway, chiming in to say that’s a pretty awesome list of personal SAS use. Also, I’d love to know if anyone has a sense for whether R is gaining traction in the corporate world/government. SAS set our small business back more than $80k this year. It’s a great program, but that’s an awfully big price tag. If R is likely to become a viable alternative for government deliverables, I’d like to start moving in that direction.
I use SAS a lot. I have taken R classes, but haven’t used it very much. I have used SPlus, which is a descendant of R.
When I read the thread title, I thought that I would suggest going with a well known package like SAS JMP, Statistica, or MiniTab. Looking at what you want to do with it though, I would suggest taking a look at R. It is free, there are people who make packages that can be used for specific areas, it is relatively easy to learn (compared to SAS at least), and it can do a lot of things. It is very quirky in its syntax at times, but it can be figured out. It should only take a short time to figure out if it can be useful to you.
I also think that trying R is your best choice. I think a SAS license is a lot more than a couple thousand a year.
One of my nephews is a newly-minted economics PhD who, among other customers, does US government funded research. 100% of what he sends back to the paying customer and to the various academic journals is done in R. Submitting raw data and R source code is *de riguer *these days.
He thinks of SAS about like we gray-beards think of QBASIC: How quaint*.
Not to say that his anecdote is the be-all and end-all of data reduction. But if I had to bet on whether R is gaining market- and mind-share I’d bet heavily on “yes.”
If you leave the “i” out of “quaint” as I did the first time you get “quant”. There’s a pretty good pun in there somewhere.
I know nothing about SAS or R programming, I just dropped in to point out that no word in the English language rhymes with pseud-orange. Really, you can look it up.
I think that R is gaining ground in areas that are not heavily regulated. I’m in pharma, and the FDA has pretty much mandated that SAS be used in a lot of areas.
I think that the main issue with R is in version control. Since a lot of R programs load in a bunch of modules developed by other people, it becomes difficult to keep track of what version of the module was used in the program. And if someone needs to recreate what happened, they have to try to figure out what versions were used or have to find the correct version. And some module developers are not that good at documenting when the module changes.
If you’re going to retire soon, then it probably isn’t that important to learn it. In my experience, folks who are proficient SAS struggle with learning R because the syntax is so different. It may be more frustrating than what it’s worth. Though I suppose learning it can’t hurt.
But if you were just starting out in your career, then I’d strongly recommend you learn it. Especially if you’re interested in going into government. Yeah yeah, open source programming can be scary if you’re dealing with regulatory work (which is what I do). But the basic packages are all well vetted, and most data analyses for the public sector doesn’t require anything that highly specialized anyway. All the training courses I’ve taken in R have been conducted or sponsored by government agencies, because all the assignments they have us doing involve running R code. Given the hefty price tag of our SAS licenses and the steady increase of R-users, I just can’t see it lasting very long in my workplace.
Duly noted. After all, the “cards;” statement in SAS assumes that inline data will be delivered in the form of punched paper cards, hardly a current way of expressing oneself. SAS leans toward being procedure oriented, though not entirely.
Question, though – I gather R is strongly object oriented. When I learned C# (and also used Java and C++ a little), I liked that a lot. So would you say R syntax is like these languages?
If it’s like Scheme, though, screw that. Counting all those sequential right parentheses is downright hard.
It might be possible to continue access to SAS via a university: join as a student and take occasional courses/seminars. But you would only do this if you enjoyed the courses–as it would probably be as expensive (or more) as a personal license.
I’m a long time SAS user. There’s almost zero support for R where I work, so it’s not really practical to use. But I used it a bit at my previous job and think it’s cool, I liked it a lot. I think you’d have a lot of fun creating the projects that you have in mind.
I didn’t have a hard time finding tips and tricks online, and since I was the only one using it, that’s all I had to go on.
We’re a Microsoft partner and there has been a big spike in demand for R. Since Microsoft purchased Revolution, they included support for R within the SQL Server 2016 as well as Power BI.
The cost of SAS is also having people consider moving to R.
