The debate over which statistical platform sits premiere over the others for data science applications rages on. The discussion often turns to the popular R and SAS environments. But to focus the dialog on performance only, a new benchmark study was just completed by commercial R provider Revolution Analytics.
The company identified a leading US marketing services provider to build predictive analytics models in 30 minutes or less, and decided to pit RRE against proprietary SAS to see which analytics platform could best meet this need. Revolution Analytics scientists utilized RRE 7 and SAS 9.4—the latest versions of both platforms—to conduct a series of analytics tasks. The results were exceptional: RRE 7 ran tasks 42 times faster than SAS, and the RRE 7 advantage ranged from 10 to 300 times the performance.
I doubt if this research will serve to tame the often heated discussion about which statistics application is better, but it could provide a more quantitative basis for the comparison. Time will tell!
Daniel, Managing Editor – insideAI News
Sign up for the free insideAI News newsletter.
Some of my SAS collegues have written a critique of the Revolution work and presented another set of benchmarks. To get another perspective, check out the SAS paper at http://support.sas.com/resources/papers/Benchmark-LASR-IMSTAT.pdf
Steve, we appreciate contrasting opinions here at insideBIGDATA, so thanks for the link to the critique.
–Daniel
It’s tough to tell how this translates to the real world when neither of you publish hardware specs. Also, wouldn’t using in-memory operations limit how big your data sets could be?