Extending the R Language to the Enterprise

Print Friendly, PDF & Email

TIBCO_meetupFIELD REPORT

Earlier this week, I attended a very informative event sponsored by the LA RUG (Los Angeles R User Group meetup) that featured the topic “Extending the R language to the enterprise with TERR & Spotfire.” Since I didn’t know much about this solution for R, I was eager to learn more. It turns out TERR and Spotfire are products of TIBCO Software, Inc. that acquired Insightful (the developer of S+, a predecessor of R) in 1987. The presenter was Lou Bajuk-Yorgan, Senior Director Product Management with the company. The presentation was polished and informative.

As I’ve reported before, I enjoy attending local Meetups to get a pulse of the industry from down in the trenches. The LARUG meetup is a great way to do this and typically attracts about 100 people for events. They usually take place at the Adconion Media Group offices in Santa Monica; they’re kind enough to provide an excellent venue, plus pizza. From past experience, I’ve learned not to arrive too late since the techies in attendance clear the pizza table like hurdlers. This time, I got there just as the food arrived. It didn’t last long, and one poor hungry looking woman got there as the last slice disappeared. It is best to consume new technology on a full stomach!

I really liked what I saw about TERR (TIBCO Enterprise Runtime for R) and Spotfire, a predictive analytics tool for non-programmers.  TERR addresses the need to make R more scalable as a production-ready environment that embraces the needs of Big Data. From what I saw, TIBCO achieved these goals and more. TIBCO’s new R architecture was built from the ground up so it solves many of R’s inherent limitations such as memory inefficiency. You can develop solutions in R and deploy in TERR with no recoding/retesting required. Speed is a virtue for TERR. We saw a benchmark test for model fitting 5 million rows of data. R took 107 seconds, and TERR only needed 17. Model scoring of 20 million rows was even more lopsided with R needing 84 seconds and TERR just 1 second. For big data implementations you can deploy TERR on the TIBCO Cloud Compute Grid, great for pure computational work. TERR integrates with both Spotfire and RStudio.

I’m anxious to give TERR a run through some big data exercises. Fortunately, the company offers a free full-featured TERR Developer Edition that you can download HERE. Check this out and after you do I think you may wish to add TERR to your big data bag of tricks.

Daniel – Managing Editor, insideBIGDATA

 

Sign up for the free insideBIGDATA newsletter.

 

 

 

 

 

 

 

Speak Your Mind

*

Comments

  1. Daniel

    Glad you enjoyed the talk, and I appreciate the article. Happy to answer any questions you or your readers have.

    Lou