06-29-2017 08:32 AM
There is no better place to ask this question than here!
I have been using SAS for a couple of years now, and i love it. The learning curve is steep but the procedures are well described and standerized, tested and reliable.
Macro language is a big plus
I have recently starting coding with R, to test the environment. Som of my friends praised the flexibility and diversity of R, especially the graphics part.
I guess it takes time to get used to R but i find it fragmented, with many ways to do the same thing (depending on the library you choose) which becomes confusing
People used to complain SAS is expensive, but with the introduction of SAS studio, which is not only free but web based making it very portable, I wonder if there is a reason to learn R at all?
R seems to be more resourses intensive than SAS too...
I would love to hear comments from everyone here on this subject
06-29-2017 08:54 AM
Since you are asking this queston in a SAS forum, you may have to take into account some bias towards the "SAS-is-better answer"
For me, though it does not hurt to know a bit of R/MATLAB/Python, SAS is the way to go. If you haven't already, then check out the IML language, which is a matrix based language like R.
Btw, be aware that you can call functions in the R language using IML as described here
06-29-2017 08:54 AM
First of all, SAS Studio is not free at all. It requires a SAS server backend with considerable license costs.
The only "free" SAS thing is SAS UE, which may only be used for learning purposes.
R, on the other hand, is free as in beer and free as in speech (GPL version 3).
Your complaint about the fragmentation of modules in R is valid, and it is typical for such environments. I read a similar complaint from @Reeza in one thread here.
See a similar thing happening with a new programming language here: Rust and the limits of swarm design
Will it make sense to learn R? For someone who intends to work in the analytics field for most of his/her professional career and is not yet close to retirement, absolutely.
R, IMHO, will be a major player in statistics; but its weakness is where the strength of SAS lies: SAS is a data warehousing framework that encompasses all aspects of the field, from data extraction to analytics. R does not provide anything like the metadata server, and it relies on third-party tools like SQL for ETL purposes.
As the data engineer in our analytics chain, not a single piece of my personal work could be solved with R. But the analysts will sooner or later start to include R in their analytics arsenal, but still use the SAS data warehouse for data preparation.
06-29-2017 09:53 AM
Its a non-starter of a question really. There isn't a SAS vs R comparitor as such. The real question is what is your requirements/level/needs/wants/system requirements, IT support etc. Try to analyse it more as a process, what is the process, what do you as a person have to put into that process, what technical platforms support can go into the process, what outputs are required from thee process, what technical platforms need to be supported at the end. Then identify tools that can be added to the process which can either simplify, standardise, or otherwise aid the process. In this way you create a stable working process. Starting with is XYZ better then ABC just leads to circular discussions between fans of one software or another, and no relation to your environment, or probably what you want to get out of it.
06-29-2017 10:58 AM
R is getting better with the packages but it's still a minefield and if you're using anything production you're definitely required to test it thoroughly.
R is open source, but that doesn't mean it's free. If you need a server you still need a server and if your data gets big fast you'll have issues with R. I still really like the ability to run the same code on a dataset that's 10,000 obs as well as 10,000,000 obs. The computers time is cheap, mine as a programmer is not.
There's also training and support which is non-existant unless you're paying a private company.