by Joseph Rickert
Last week in a webinar, Burtch Works, an Illinois based executive recruiting firm that specializes in finding analytic talent, released the results of their third annual survey of "quantitative business professionals". Other than having attended this webinar, I have no knowledge of Burtch Works, but I am willing to take their survey at face value as a good faith, objective effort to answer a question that must be of some importance to a firm in the business of recruiting data scientists and other quantitative professionals.
This year they asked just one question: Which do you prefer to use: SAS, R, or Python? - Not the question I would have thought to ask: I would not have boldly asked people to pick a winner without providing some context. There is no nuance here. Nevertheless, it is a practical, straightforward question that has provided some interesting results. The simple tally from over 1,000 responses came down like this.
Open source tools dominate overall. SAS did well with professionals with over 16 years of experience while those with less than 5 years of experience preferred R. R was also the dominant choice of analytics professionals with Ph.D's and Master's Degrees.
In an original, and I think revealing attempt to distinguish job function, the Burtch team distinguishes between Data Scientists and Predictive Analytics professionals. This unusual dichotomy led to the following chart.
I interpret the left side to say that programming professionals doing data wrangling overwhelmingly prefer the opensource languages. It seems reasonable that the Python could have the edge here over R, and that very little of the pie goes to SAS. On the right, it is clear that R and SAS are the tools preferred by people building statistical models. These charts make good sense, and it is very nice to see R doing well in both categories.
The Burtch folks also slice their data by region and industry. You can review all of their findings by watching the short video above, or by examining the report posted on the Burtch website. The report concludes with a few by cherry picked comments from the survey responders. My favorite is: R – Anytime, Anywhere, Anything.
Hmm... I don't think the numbers quoted justify the headline for the post. "SAS losing ground to open source tools" would be nearer the mark, as 43% of modelers prefering SAS is hardly "failing". It's also not surprising that young data scientists are using open source languages a lot, given that so much big data storage is on open source platforms and they can install and use them for free.
Posted by: Steve M. | July 31, 2016 at 03:45
"Open source tools dominate overall. SAS did well with professionals with over 16 years of experience while those with less than 5 years of experience preferred R. R was also the dominant choice of analytics professionals with Ph.D's and Master's Degrees. "
On the surface, this does not surprise me. Professionals with over 16 years experience would indicate to me those who are employed by corporations (worldwide?), and consultants who grew up with the less heavyweights of SPSS, Minitab and other less powerful software, but migrated to SAS when they entered the business environment as opposed to academic world.
Those with less than 5 years of experience and analytics professionals with Ph.D's and Master's Degrees preferred R also does not surprise me. Academics (Ph.D's and Master's Degrees) are always looking for the least expensive investment, in time of learning curve and monetary cost to facilitate their studies. Those with less than 5 years of experience have probably not been exposed to a corporate environment that requires a very robust and fully supported software package that is known for its reliability.
The paradigm may well be shifting away from SAS, but SAS is hardly failing..
Posted by: David Sellar | August 01, 2016 at 14:07
SAS is hardly failing if it made the top list of this article recently
http://time.com/money/4328180/most-valuable-career-skills/
Posted by: J S | August 02, 2016 at 06:51