A quick follow-up on my previous email:

>> On 25 March 2012 03:49, Olivier Grisel <[email protected]> wrote:
>>> [..]
>>>
>>> Another way to rephrase that question: what is the typical sweet spot
>>> for the dataset shape when doing classification Gradient Boosted
>>> Trees? What are reasonable values for the number of estimators in
>>> various application domains?

GBRT is relatively robust w.r.t. overfitting if you use shrinkage (
see the docs). The choice of the number of estimators is mainly
governed by the available computational resources. As usual you have
to balance between training vs. test time performance - if you don't
care about test time performance you should use as much base learners
as you can afford**. If test time performance matters (e.g. latency
constraints in a web app) things look different :-)

** I remember that a student of mine once used more than 10K base
learners for the yahoo learning to rank competition.

>>>
>>> --
>>> Olivier
>>> http://twitter.com/ogrisel - http://github.com/ogrisel
>>>
>>> ------------------------------------------------------------------------------
>>> This SF email is sponsosred by:
>>> Try Windows Azure free for 90 days Click Here
>>> http://p.sf.net/sfu/sfd2d-msazure
>>> _______________________________________________
>>> Scikit-learn-general mailing list
>>> [email protected]
>>> https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
>>
>> ------------------------------------------------------------------------------
>> This SF email is sponsosred by:
>> Try Windows Azure free for 90 days Click Here
>> http://p.sf.net/sfu/sfd2d-msazure
>> _______________________________________________
>> Scikit-learn-general mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
>
>
>
> --
> Peter Prettenhofer



-- 
Peter Prettenhofer

------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here 
http://p.sf.net/sfu/sfd2d-msazure
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to