2013/7/8 Josh Wasserstein ribonucle...@gmail.com:
Thank you Lars. I didn't see any deprecation warnings. Also, from what I can
tell, one of the primary examples for model selection in the documentation
uses cv as an argument to clf.fit:
clf = GridSearchCV(SVC(C=1), tuned_parameters,
2013/7/4 Josh Wasserstein ribonucle...@gmail.com:
I am confused, what exactly is deprecated? Was there anything in the code I
sent in my emails that is deprecated?
Didn't you get a deprecation warning?
--
Lars Buitinck
Scientific programmer, ILPS
University of Amsterdam
On 07/03/2013 09:54 PM, Vlad Niculae wrote:
Also, it's not that GridSearch is sensitive in itself, but remember
you're doing LeaveOneOut, so for every grid point you are actually
doing `n_samples` calls to clf.fit.
Maybe one of these calls is significantly slower than others due to scaling.
2013/7/4 Andreas Mueller amuel...@ais.uni-bonn.de:
Why is there a cv option to fit? That is deprecated, right?
Turns out we forget this one.
--
Lars Buitinck
Scientific programmer, ILPS
University of Amsterdam
--
This
2013/7/4 Lars Buitinck l.j.buiti...@uva.nl:
2013/7/4 Andreas Mueller amuel...@ais.uni-bonn.de:
Why is there a cv option to fit? That is deprecated, right?
Turns out we forget this one.
No, wait, I was jumping to conclusions. It is deprecated, to be
removed in 0.15. fit parameters to
On 07/04/2013 12:16 PM, Lars Buitinck wrote:
2013/7/4 Lars Buitinck l.j.buiti...@uva.nl:
2013/7/4 Andreas Mueller amuel...@ais.uni-bonn.de:
Why is there a cv option to fit? That is deprecated, right?
Turns out we forget this one.
No, wait, I was jumping to conclusions. It is deprecated, to be
On 07/04/2013 10:38 PM, Josh Wasserstein wrote:
I am confused, what exactly is deprecated? Was there anything in
the code I sent in my emails that is deprecated?
Yes. Passing a cross-validation class to the fit method of grid search.
This was just ignored. You should have passed it to
Thank you Vlad. I think you are right and there may be a problem with
parallel jobs.
When I run the code with the verbosity option enabled I see output coming
out slowly. The strange thing is that doing a simple SVM fit is basically
instantaneous (literally less than half a second), so I am not
Hmm, I noticed that if I run
from sklearn import preprocessing
X = preprocessing.scale(X)
beforehand, it runs extremely fast!
Why is that?
Jacob
On Wed, Jul 3, 2013 at 3:07 PM, Josh Wasserstein ribonucle...@gmail.comwrote:
Thank you Vlad. I think you are right and there may be a problem
Perhaps more oddly, why is GridSearchCV so sensitive to it (note that a
simple svm.SVC().fit(X,y) without scaling was already fast.
In other words, it looks like scaling affects GridSearchCV in particular.
Jacob
On Wed, Jul 3, 2013 at 3:35 PM, Josh Wasserstein ribonucle...@gmail.comwrote:
2013/7/3 Josh Wasserstein ribonucle...@gmail.com:
Hmm, I noticed that if I run
from sklearn import preprocessing
X = preprocessing.scale(X)
beforehand, it runs extremely fast!
Why is that?
Because support vector machines are quite sensitive to extreme feature
values. You should always
Also, it's not that GridSearch is sensitive in itself, but remember
you're doing LeaveOneOut, so for every grid point you are actually
doing `n_samples` calls to clf.fit.
Maybe one of these calls is significantly slower than others due to scaling.
On Wed, Jul 3, 2013 at 10:42 PM, Lars Buitinck
2013/7/3 Josh Wasserstein ribonucle...@gmail.com:
Perhaps more oddly, why is GridSearchCV so sensitive to it (note that a
simple svm.SVC().fit(X,y) without scaling was already fast.
In other words, it looks like scaling affects GridSearchCV in particular.
According to your logs, it's slow
13 matches
Mail list logo
