SAS Versus R

13 thoughts on “SAS Versus R”

1. Interesting Blog. Globally, SAS is still the market leader in available corporate jobs but ‘R’ has also been reported to increase over last few years. What do you think about it ?

2. I would bet money that every company that you have listed as SAS users also have R users on staff, and every single company you have listed as R users has SAS users.

   1. Excellent point. And true point for Uber, Ford, Bing and many many others 🙂

3. I have used both SAS and R for many years.
   SAS has great data handling capabilities. If you have complicated transaction data and need to heavily process it then SAS works well and is much more flexible than standard SQL. If the new Hadoop version of SAS works well (my company is testing it out), I think SAS will remain a strong power in corporate processing.
   However SAS has some weaknesses.
   1. It is very expensive.
   2. Graphics are complicated and fiddly. I think the JMP program from SAS has better capabilities but I have never used it.
   3. SAS’s machine learning algorithms are mediocre and are about 10 years behind the state of the art, I think SAS needs to buy a machine learning company. For heavy machine learning algorithms my company uses Salford Systems tool which are easy to use and not super expensive.

   There are many things I like about R. The algorithms are powerful and usually easy to use even for an inexperienced R user. Second it is open source, R has some weaknesses though especially in a corporate environment.
   1. IT security hates open source software especially on a server that has sensitive data. Therefore a company will very likely have to purchase Microsoft’s (Revolution) R which is still less expensive than SAS. This also means that you won’t have access to state of the art algorithms until Microsoft implements them.
   2. R’s data handling is complicated and difficult for complex data. Usually most users will do most of the data crunching in Hive or some other SQL like program and then read the processed file into R. The way I use R is to embed it in a SAS program and then use the X command and SAS file pipes to control the R processing

4. I have used SAS 10+ years and R 5+ years.

   Main Pros of SAS :
   – Good GUI (SAS Enterprise Miner), however You have to be SAS and R expert in order to use it fully (SAS coding, R coding etc)
   – Special solutions like Credit Scoring and Survival Data Mining. Free R and Microsoft R currently lacking such good solutions (there are some R packages like discSurv, but it takes time to learn them).
   – Very good support by SAS (technical support, good SAS courses and books, SAS Global Forum etc)

   Main Cons of SAS:
   – SAS programming language is somewhat inflexible and dated (especially SAS macro language, DATA step)

   Main Pros of R:
   – Good frameworks for automated data mining (R packages mlr, caret)
   – Very good and flexible language for programming
   – State-of-arte algorithms ready for use immediately after publication (> 10 000 packages!)
   – Reusage of R solutions / functions within SAS, Vertica etc

   Main Cons of R:
   – No good GUI like SAS Enterprise Miner
   – Lack of special solutions like credit scoring, survival data mining

5. I think it important to note when looking at the above comparison that each of the companies listed as employing R are also heavy SAS users.

   Analytics does not have to exist in an SAS OR R world, the market leading organisations are embracing SAS AND R. Using silo’d technology is no different than having silos of data that you can’t get the most out of.

6. The statement that SAS doesn’t have great graphical capabilities is wrong. With SAS Visual Analytics and gplots everything is possible.

   1. But you have to pay also for it and that is not cheap 😉

7. Because R is open source, it has now been integrated in nearly every major analytic tool (and several not so major ones) on the planet. I have personally programmed R from within SAS, Matlab, Alteryx, Tableau, Anaconda Python, and Spark and there are many more. The entire question of R versus SAS – or even R versus nearly *anything* – is broken. Analytics and data science to means R And SAS And Python And Java And web pages And And And And And….

   Lots of good commnts here on relative strengths. From my personnal, narrow perspective I would echo SAS for big data, smooth commercial operation, and specialized tools for different industries – financial, medical, and many others. SAS’s great strength in regulatory environments hasn’t been mentioned until now but it’s a huge factor in those industries where models get sued. R for open source, latest developments, and, well, integration with many other tools. That’s really the point of this post. Build R into whatever proprietary system you have and take advantage of its strengths. R integrates so widely and so well that debates on R *versus* anything should be a thing of the past.

8. This comparison only scratches the surface. Very, very gently. So many things are missing, don’t even know where to start… Look at Quora, where comprehensive comparisons exist. SAS and R are highly comparable in terms of statistical methods implemented. They compete strongly. SAS graphing module is very powerful if one knows how to use it. Price? In most cases – nobody cares, really. If a company uses SAS, in 99.999% its income is much greater than the cost of SAS licence. which becomes negligible. And I say this as a big fan of R, using R for the last 15 years. It’s not mentioned, that SAS is validated while R not – this makes problems in controlled environments, like clinical research. Luckily it’s not a big challenge to build a fully validated environment based on R and many companies do that. It’s a myth, that R cannot be used for that purpose. Many clinical research companies, including the top most ones employ R in their processes. Other thing – R is much more flexible than SAS. One can embed R even on simplified computers, like BananaPi or Odroid. Don’t try this with SAS. This way one can make a fully independent computing unit with own user interface. One more thing: Bing, Ford and Uber belong to the group of the most well known R users, for years.

9. I have been using SAS for about 12 years and I love it. With that said, I still prefer to create my visualizations in Tableau.

10. The very important topic discussed artistically

11. As a Newbie, I am always browsing online for articles that can help me. Thank you

Comments are closed.