[R] Are you experienced in SAS and R as well? Which of these would you recommend?

Fri Nov 23 16:36:49 CET 2001

Michael,

I know something of SAS and a little of R.
I would think the big question is, "what do you intend using it for?"
Both R and SAS are interpreated stats computing languages. R is a great
stats tool  but I don't know about it's data management side where SAS is v
good at handling large datasets (I'm talking of 1-10M records by 100 vars)
not small stuff of 1000 obs by 5 or 10 vars.

SAS has a lot of built-in procedure that are easy to understand and follow
such as proc means/summary; the relative novice will get to grips with this
fairly quickly and be able to do basic but useful things.  The equivalent
functions in R are the "apply" variants and require a bit of getting used
to. Also, the editing features in PC-SAS are v. friendly and pleasant to
use. This (in my view) is in contrast to R where I have to use NOTEPAD (in
Windows anyway) to create a text file which is then included into the
console (I find it a bit cumbersome). SAS also has really good
documentation that explains a lot whereas with R you often have to be a bit
more expert.

Where R comes into its own is when things get a bit complicated - there's
lots of packages available on CRAN which do wonderful things that SAS
doesn't. In SAS you'll have to resort to MACROS and IML and even then the
building blocks won't always be available to you. Yes, SAS has proc's for
GLM and maximising nonlinear functions etc. which are not to hard to apply
but once you try to get outside the rigid SAS framework life gets difficult
- I once tried to put together a Neural Net with Proc NLIN and eventually
gave up because the of all the messy programming that had to be done -
mabye some day in another life I'll get back to it!

Even though R has great statistical functionality it's main drawback (to me
at any rate) is that most of the underlying code is written in C or
Fortran. Not a problem if you only require to work with R packages as they
stand. But if you want to get your hands dirty chances are you'll have to
hack someone else's C or Fortran code - and I have to say that some of the
Fortran I've seen has been fairly unintelligible. This makes it
particularly hard to make anything but rudimentary changes to the code
which then must be compiled and loaded into your R program.

The last point worth mentioning is that R, as I understand it, is designed
to work on vectors (that's why the "apply" function is so important). It
doesn't like loops so that any code involving large complex looping
constructs is better written in C or Fortran and loaded in. Loops in SAS on
the other hand don't seem to be too much of an issue.

Anyway I hope this helps. My advice is do what I tend to do - try to get
your dept or company to spend the $150 on SAS and get R free. You'll then
get be able to get answers using both tools. It's a little bit inefficient
but often the insight gained looking at a task from different viewpoints
helps the understanding and final product/output a lot better.

All the best,

Gerard

                    Michael Reinecke                                                                                   
                    <mreineck at uos.de>           To:     r-help at lists.R-project.org                                     
                    Sent by:                    cc:                                                                    
                    owner-r-help at stat.ma        Subject:     [R] Are you experienced in SAS and R as well? Which of    
                    th.ethz.ch                  these would you recommend?                                             

                    23/11/01 11:31                                                                                     
                    Please respond to                                                                                  
                    Michael Reinecke                                                                                   

Hello!

Disapointed about SPSS I have to choose another statistic program. And
altough I sympathise with the idea of a non-commercial software-project
like R and I like the spirit of the R community (and of course I am not
keen on paying 150$ to SAS for a one-year students license), I will
probably buy SAS, because people I will work with use SAS and I want a
close cooperation with them. Besides I guess that SAS is still much more
powerful than SAS.

But: can ´t you convince me to choose R? I would appreciate that very much,
because my sympathy is with the r-project. Maybe the syntax of both
programs is quite similar, so I will still be able to exchange a lot of
experience with my SAS-using future collegues!? Or I could even convince
the newcomers to swap to R!?

If you know some arguments, I should consider for my decission, I am
looking forward to hear (or read) of you!

Michael

Michael Reinecke
Turnerstr. 3a
49076 Osnabrück
Germany
Tel: +49-541-64147
mreineck at uos.de

The information in this email, and any attachments transmitted with it, are confidential 
and are for the intended recipient only. If you receive this message in error, please 
notify us via postmaster at cso.ie.

To see the latest figures from the CSO go to http://www.cso.ie
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._