Opinions on the R/SAS squable. 1) SAS can not and does not have the intellectual firepower that R does. R has thousands of Statistical/machine learning professors/lecturers/Post docs developers around the world. This keeps ideas and techniques flowing. Whether there all correct or useful is another matter. No company can keep up with those resources. 2) Obviously SAS has some excellent statisticians/machine learning developing SAS code. But by not having the sheer weight of numbers of quality people examining and assessing the ideas, the techniques can become outmoded. Rejected techniques like CHAID are still maintained. Peer review is importand and R has more people at the highest level who can peer review openly and sometimes are quite nasty. 3) Every vendor has a way of intgegrating different platforms and data streams together. R can access the power of a teradata warehouse. R can be implemented on a mainframe thru code generation of a R model into a DB2/COBOL program. SAS does something similar with its model. Integration of R and data across an organisation business model is pretty routine. 4) I think SAS future is as a statistical service company. They are not a leader in statisitical /machine learning research(more of a follower), ETL can be done by so many different platforms, deployment of R models across platforms is becoming easier. But I think they do statistical services well.
... View more