## Why R (and not SAS) again.

Ten years ago, when I first came to INRA, I learned SAS. This was THE statistical analysis language. Unfortunately, SAS was not really affordable (this is still true), so back to school, if I wanted to run some analysis, I had to take appointments with my professor who was the only person with SAS on its computer. Not a very convenient solution to move forward.

Thus, I tried to figure out if I had an alternative to SAS…that’s the way I discovered R ! I’ve already acknowledged Phillipe Besse (who by the time offer both SAS and R scripts for data mining on his website), after a friend Ph’D defense, but I ‘d like to repeat my self thanks to him and the R community, I really learned a very valuable skill for my career.

Once fully convinced that R was a good solution (and SAS might be a problem), I tried to convince my colleague that moving from SAS to R could be worthwhile on the long term. I unfortunately received the same old song : SAS was very popular and all kind of not so bullet proofed arguments. With time, using R became more and more obvious, so that I fortunately had less and less often to argue about why R and not SAS. In fact :

- Most students have now basics skills in R (and less often in SAS)
- R implement most of the newly published method
- R is pretty versatile (though not perfect in every area)

The excellent freakonometrics just published a presentation , presenting R and adding a small review of R popularity. The licence fees are pretty self explanatory, and I hope that referring to them will give me more credits, next time I’ll try to convince people that the cost/benefits ratio of SAS, is questionable.