Statistical and predictive analytics software is fast becoming a next big competitive landscape in the business intelligence market. IBM recently purchased venerable 40 year old vendor SPSS and will undoubtedly use it's muscle to expand the SPSS footprint. SPSS predictive models are already embedded in a host of business applications. Expect that development to escalate. Expect also tighter integration with IBM-owned BI platform Cognos to provide cradle-to-grave intelligence.
Revolutions recently announced a shake-up at REvolution Computing, to include a generous new round of venture funding coupled with a replacement senior management team. The much-needed new CEO is none other than Norman Nie, currently a professor at Stanford, but an SPSS founder back in 1968. Nie, ironically, recently settled with SPSS in a trademark dispute that threatened to derail the IBM acquisition. It's a small statistical world.
Nie seems a perfect fit for the challenges currently facing REvolution as it attempts to build a profitable R business model. An accomplished political scientist, Nie should resonate well with an R community that's significantly academic at present. Hopefully, he'll introduce R to the BI world much as he did SPSS many years ago. Perhaps REvolution R Enterprise, REvolution's paid subscription version of R, can fill the ease-of-use role that SPSS served for many years. SPSS is probably more popular among non-statisticians such as marketers, psychologists and business strategists than it is among hard-core stats types, it's easy-to-use Windows interface a differentiator with that segment. Though I haven't seen it, I've been told the R Productivity Environment included in Enterprise is a slick user interface and development tool that insulates casual users from the tedium of writing R code. That's certainly a good start, as is REvolution's obsession with performance.
REvolution inherits much goodwill as the commercial vendor of R. R's meteoric rise in academic and research worlds is nothing short of remarkable. Even with an estimated 2M worldwide users currently, R's popularity continues to rise. And with an easily extensible object-oriented architecture and so many zealous users, it's little wonder that R laps the competitive field incorporating the latest statistical techniques. R innovation is quite astonishing, testimony to the power of the open source software model. The number of freely-available packages developed by the R community is approaching 2000. Take a look at the accessible R machine learning modules. Now check how many of these techniques are implemented in other statistical platforms.
Finally, browse the excellent Revolutions blog, assembled by dynamic VP of Community, Dave Smith. The blog entries cover a wide range of REvolution Computing, R and other statistical/predictive analytics topics. Dig into the Categories and other links. Smith's done a great job providing lots of interesting reading, balancing the needs of different constituents. His community work is critical to bridging the academic-business gap for REvolution.
SAS, of course, is still the leader in the statistical market. A private $2.5B/year revenue company, SAS has long been the envy of statistical and BI vendors for its persistent pricing power in the market. And for enterprise, big-data analytics, SAS is still king. The company, though, seems vulnerable on a number of fronts.
One safe bet is that IBM, with newly-acquired SPSS and Cognos, is gearing up to take on SAS in the high-end enterprise analytics market that features very large data and operational analytics with significant capacity challenges. In this segment, IBM can leverage its hardware, database and consulting strengths to become a formidable SAS competitor. A strong statistical and quantitative focus is central to IBM's Smarter Planet strategy.
A major enabler of SAS's ascendance in the early 80's was the enthusiastic following of statisticians emerging from top schools who'd learned SAS as part of their curricula. When those loyalists entered the work world, their platform of choice for statistical work was, not surprisingly, SAS. Indeed many top SAS analysts have used the product for more than 20 years, often turning to SAS for mundane data and programming needs as well. SAS provided one stop shopping for many.
In 2009, the statistical platform of choice at top graduate schools around the world is R, not SAS. And like their SAS brethren of 25 years ago, as they enter the work world, these new statisticians wish to continue using what they've learned and love -- only now it's R. This transition from SAS to R in academy is, I believe, a leading indicator of change in the statistical marketplace 5 to 10 years out.
A number of start-up companies promoting competitive SAS language tools at a fraction of SAS prices may begin chipping away at many SAS annuity customers. As I wrote in last week's blog, WPS from World Programming Systems is an outstanding SAS compiler that can replace expensive SAS licenses in many cases – especially those primarily used for data step programs. Similarly, another competitor, Carolina, from Dulles Research, LLC, converts Base SAS program code to Java, which can then be deployed in a Java run-time environment. Large SAS customer Discover Card is currently evaluating Carolina as a replacement for some its SAS applications. A larger combination of inexpensive WPS or Carolina and R might make sense for other shops seeking to cut SAS licensing costs even more dramatically. If I'm a CIO, an evaluation of these product permutations is a no-brainer.
My expectation is that this heightened competition will be a boon for statistical consumers of BI. I anxiously await developments over the coming months.
Steve Miller's blog can also be found at miller.openbi.com.