Decision trees in R with big data

Revolution R Enterprise, a commercial distribution of R, includes
external memory algorithm implementations of both decision trees and
decision forests. These are geared for "tall" data--the two million
rows wouldn't be a problem, nor would two billion, but the 20,000
attributes would probably challenge them.

It's probably worth a look (and is available for free for academic use):

www.revolutionanalytics.com

Hope this helps!

--Rich Calaway

Hi,

I have successfully used rpart but with a few thousands rows, and a few
hundred input attributes. When using data with ~2 million rows (instances),
and ~20,000 input attributes (typical data sizes in my application), I get
memory problems when using rpart.

Does anyone know of a Decision tree algorithm that works in R with big
data?

Thanks!

        [[alternative HTML version deleted]]

_______________________________________________
R-sig-hpc mailing list
R-sig-hpc at r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-hpc

Rich Calaway
Documentation Manager
Revolution Analytics, Inc.
1505 Westlake Ave North Suite 520
Seattle, WA 98109
richcalaway at revolutionanalytics.com
ph: 206-456-6086 (direct line)
855-GET-REVO