[ml] decision trees / random forests?

[Mike Schachter]

Hi Glenn!

We have given a workshop on random forests in the
past, the slides can be found here:

https://docs.google.com/present/view?id=dcbvcg62_126d7xpmtf8

There is source code in the git repository that uses
random forests to do digit recognitiion. The git repo
itself can be checked out like this:

git clone git://ml-noisebridge.git.sourceforge.net/gitroot/ml-noisebridge/ml-noisebridge

You can find the random forest example in:

ml-noisebridge/workshops/randomForests_handwriting/rf_semeion.R

To use randomForests in R, you need to install the package
that can be located here:

http://cran.r-project.org/web/packages/randomForest/index.html







On Mon, Jan 2, 2012 at 2:32 PM, Glenn Wright
<infinite.perplexity at gmail.com> wrote:
> Hey all, anyone have any background with decision trees, especially (what I
> think are known as) ensemble methods such as random forests?  Straight out
> of the Stanford class I found an interesting problem to work on, but for a
> variety of reasons I think it might work better with decision trees than
> with neural networks.  Unfortunately, that's not an area we covered in the
> class.
>
> I programmed a very simple, non-ensemble implementation by hand, and while
> it's ridiculously slow, it does seem to solve the problem.  So I downloaded
> the "milk" python library, which uses NumPy, and tried running its
> (obviously much faster) random forest model.  And...well...it seems to be
> getting the right answers, but I'm getting to the point where I'd like to
> compare notes with someone before I get too far out on a limb.  Get it?  Out
> on a limb?  Trees?  Okay, I'll shut up now :)
>
> --
> -Glenn
>
> _______________________________________________
> ml mailing list
> ml at lists.noisebridge.net
> https://www.noisebridge.net/mailman/listinfo/ml
>