addressalign-toparrow-leftarrow-rightbackbellblockcalendarcameraccwchatcheckchevron-downchevron-leftchevron-rightchevron-small-downchevron-small-leftchevron-small-rightchevron-small-upchevron-upcircle-with-checkcircle-with-crosscircle-with-pluscrossdots-three-verticaleditemptyheartexporteye-with-lineeyefacebookfolderfullheartglobegmailgoogleimageimagesinstagramlinklocation-pinmagnifying-glassmailminusmoremuplabelShape 3 + Rectangle 1outlookpersonplusprice-ribbonImported LayersImported LayersImported Layersshieldstartickettrashtriangle-downtriangle-uptwitteruseryahoo

"Official"­ BARUG July 2013 Meeting


6:30 PM - pizza and networking
7:00   - announcements
7:10   - Joseph Rickert: "Statistics with Big Data: Beyond the Hype"
7:40   - Saptarshi Guha: "On Using Terra as an Extension 
Language for R: A Fast Alternative to C"

Abstract For Saptarshi's Talk

One approach to improve performance in R is to write the critical region in C or C++ and then call this from R. Packages to make this effort easier and more fluid exist such as Rcpp and inline.  However, not all R users are comfortable in either C or C++. 

Terra is a new research project developed by Zach Devito of Stanford. To quote, "Terra is a new low-level system programming language that is designed to interoperate seamlessly with the Lua programming language". Users of Terra can write Lua code (which is JIT compiled via LuaJIT) or Terra (which is compiled via LLVM)
and both can interoperate with each other.

In this talk, I demonstrate how Terra can be used as an extension
language for R with similar performance to C based extensions.

Saptarshi is a Data Scientist at Mozilla Metrics.


Join or login to comment.

  • Tory S.

    Hi, were are the presentations posted?! Wasn't able to attend this one. Thanks!

    July 31, 2013


    I am doing a research using agglomerative hierarchical clustering with R. my idea is this: out of n observations with p variables, instead of clustering n observations, i want R to do an agglomerative hierarchical cluster using complete linkage and stop the clustering at h=(n+p+1)/2 instead of the usual clustering of the whole n observation through agness or hclust. how can this be done? i mean is this possible in R?

    July 3, 2013

    • Matthew J.

      See the description of package sparcl in http://cran.r-project...­. That is the closest think I see to what you describe. But there are several other clustering packages on that page; a lot more than just agness or hclust.

      July 31, 2013

  • Dennis N.

    First time meeting with BARUG. Good talks. I was hoping for a time slot for sharing tips in using R, but perhaps that only happens at some meetings

    July 30, 2013

  • Matthew J.

    Joe's information about the three levels of data science programming was clear and valuable.

    July 30, 2013

  • Michael R.

    Need to let someone else attend tonight...should be a good one.

    July 30, 2013

  • Dr Nigel G.

    I am bringing a colleague - Ian L

    July 30, 2013

  • David W.

    Here's to healthy genomes, proteomes and all the -omes around us.

    July 24, 2013

  • Justin

    Bummer! Last minute trip

    July 24, 2013

  • Saptarshi Joy G.

    I would like to mention that the talk will focus on coding examples and not so much statistical modelling.

    July 9, 2013

Our Sponsors

People in this
Meetup are also in:

Sign up

Meetup members, Log in

By clicking "Sign up" or "Sign up using Facebook", you confirm that you accept our Terms of Service & Privacy Policy