Re: [Scikit-learn-general] decision_function for OneVsRestClassifier

Le 18 avril 2012 13:54, Ian Goodfellow a écrit : > I'm not sure if it's correct. I'm also not familiar enough with the > library to be sure that no similar functionality exists somewhere. (I > think David Warde-Farley mentioned that there was a plan to do > something like this at some point) > Can

Re: [Scikit-learn-general] decision_function for OneVsRestClassifier

I'm not sure if it's correct. I'm also not familiar enough with the library to be sure that no similar functionality exists somewhere. (I think David Warde-Farley mentioned that there was a plan to do something like this at some point) Can anyone comment on whether this is actually the decision fun

Re: [Scikit-learn-general] SVC kernel documentation

On Wed, Apr 18, 2012 at 04:09:14PM -0400, David Warde-Farley wrote: > > Do you suggest a link from the docstring of the object to the docstring > > of the module? > See my pull request: https://github.com/scikit-learn/scikit-learn/pull/783 Yes. Thanks, it is useful. > Is it frowned upon to link

Re: [Scikit-learn-general] SVC kernel documentation

On Wed, Apr 18, 2012 at 09:08:15PM +0200, Andreas Mueller wrote: > Every time "SVC" it is mentioned in the text it is a clickable link to > the docstring. > This is a consistent feature throughout all narrative docs. Oops. My bad. > > Similarly, it would be nice to have a link back to the first

Re: [Scikit-learn-general] decision_function for OneVsRestClassifier

Op 18 april 2012 18:53 heeft Ian Goodfellow het volgende geschreven: > Is there a reason why OneVsRestClassifier does not implement > decision_function? > It looks like the correct implementation is just to concatenate > together the decision_function outputs of the individual estimators_. Well,

Re: [Scikit-learn-general] SVC kernel documentation

> http://scikit-learn.sourceforge.net/stable/modules/svm.html#kernel-functions > > The dev version does list coef0 in that section. > > General comment: is there any link to the SVC docstring from the narrative > docs? I couldn't find one, had to go to the "Reference" tab. Every time "SVC" it is m

Re: [Scikit-learn-general] SVC kernel documentation

On Wed, Apr 18, 2012 at 01:26:30PM -0400, Ian Goodfellow wrote: > Would it be possible to update the SVC documentation to specify which > kernel functions are actually implemented? > For example, is 'poly' K(x,y) = (x^T y )^d or is it K(x,y) = (x^Ty + 1)^d? > Also, why are gamma and coef0 listed as

Re: [Scikit-learn-general] SVC kernel documentation

Am 18.04.2012 19:42, schrieb Ian Goodfellow: > Yes, thanks. > It might be a good idea to put something like that in the docstring. > Do you think a reference to the docs would suffice or do you think the whole thing should be listed there? -

Re: [Scikit-learn-general] SVC kernel documentation

Yes, thanks. It might be a good idea to put something like that in the docstring. On Wed, Apr 18, 2012 at 1:29 PM, Andreas Mueller wrote: > Hi Ian. > Is that what you are after: > http://scikit-learn.org/dev/modules/svm.html#kernel-functions > Cheers, > Andy > > Am 18.04.2012 19:26, schrieb Ian G

Re: [Scikit-learn-general] sparse.OneClassSVM usage problem

On 04/18/2012 06:12 PM, Olivier Grisel wrote: > Le 18 avril 2012 06:51, Dimitrios Pritsos a écrit : >> Ok I ve created the Issue in Git ( I Guess this is the proper place to do >> what Olivier suggested me in the last email in this list) >> >> however what is CV an where I can enable indices=True,

Re: [Scikit-learn-general] SVC kernel documentation

Hi Ian. Is that what you are after: http://scikit-learn.org/dev/modules/svm.html#kernel-functions Cheers, Andy Am 18.04.2012 19:26, schrieb Ian Goodfellow: > Would it be possible to update the SVC documentation to specify which > kernel functions are actually implemented? > For example, is 'poly'

[Scikit-learn-general] SVC kernel documentation

Would it be possible to update the SVC documentation to specify which kernel functions are actually implemented? For example, is 'poly' K(x,y) = (x^T y )^d or is it K(x,y) = (x^Ty + 1)^d? Also, why are gamma and coef0 listed as relevant to the poly kernel? I don't know of anything in the literature

[Scikit-learn-general] decision_function for OneVsRestClassifier

Is there a reason why OneVsRestClassifier does not implement decision_function? It looks like the correct implementation is just to concatenate together the decision_function outputs of the individual estimators_. The current setup is kind of frustrating to use since the interfaces of LinearSVC an

Re: [Scikit-learn-general] sparse.OneClassSVM usage problem

The *problem is restricted to Sparse Matrices* for dense Matrix everything is working fine. On 04/18/2012 04:51 PM, Dimitrios Pritsos wrote: Ok I ve created the Issue in Git ( I Guess this is the proper place to do what Olivier suggested me in the last email in this list) however what i

Re: [Scikit-learn-general] non-exhaustive grid search

Le 18 avril 2012 07:35, Andreas Mueller a écrit : > Am 18.04.2012 16:25, schrieb Mathieu Blondel: >> Hello, >> >> Recently, I had to learn a classifier with 3 hyperparameters (each >> taking several possible values)  and an exhaustive grid search was >> really expensive. However, in some situation

Re: [Scikit-learn-general] sparse.OneClassSVM usage problem

Le 18 avril 2012 06:51, Dimitrios Pritsos a écrit : > > Ok I ve created the Issue in Git ( I Guess this is the proper place to do > what Olivier suggested me in the last email in this list) > > however what is CV an where I can enable indices=True, as it is suggested to > the following message: I

Re: [Scikit-learn-general] non-exhaustive grid search

Am 18.04.2012 16:25, schrieb Mathieu Blondel: > Hello, > > Recently, I had to learn a classifier with 3 hyperparameters (each > taking several possible values) and an exhaustive grid search was > really expensive. However, in some situations, some hyperparameters > are more important than other

[Scikit-learn-general] non-exhaustive grid search

Hello, Recently, I had to learn a classifier with 3 hyperparameters (each taking several possible values) and an exhaustive grid search was really expensive. However, in some situations, some hyperparameters are more important than others. For example, for a SVC with RBF kernel, my experience is

Re: [Scikit-learn-general] where is the feature_extraction.text.WordNGramAnalyzer

Le 18 avril 2012 06:54, 张鹏 a écrit : > hi  all >     I can not found feature_extraction.text.WordNGramAnalyzer in 0.11 . Has > this class been removed ? Yes, please read the changelog of the development version here: http://scikit-learn.org/dev/whats_new.html The new documentation for the text

Re: [Scikit-learn-general] Video of Jake's tutorial at PyData

Cool, I'll have a look tonight! Jaques 2012/4/18 Mathieu Blondel > > > On Wed, Apr 18, 2012 at 4:51 AM, Olivier Grisel > wrote: > >> >> http://marakana.com/s/scikit-learn_machine_learning_in_python,1152/index.html >> >> > Good job, Jake! > > Mathieu > > > --

[Scikit-learn-general] where is the feature_extraction.text.WordNGramAnalyzer

hi all I can not found feature_extraction.text.WordNGramAnalyzerin 0.11 . Has this class been removed ? -- [email protected] RDC

Re: [Scikit-learn-general] sparse.OneClassSVM usage problem

Ok I ve created the Issue in Git ( I Guess this is the proper place to do what Olivier suggested me in the last email in this list) however what is CV an where I can enable indices=True, as it is suggested to the following message: ValueError: X and y have incompatible shapes: (180, 255) vs

Re: [Scikit-learn-general] Video of Jake's tutorial at PyData

On Wed, Apr 18, 2012 at 4:51 AM, Olivier Grisel wrote: > > http://marakana.com/s/scikit-learn_machine_learning_in_python,1152/index.html > > Good job, Jake! Mathieu -- Better than sec? Nothing is better than sec when it c

Re: [Scikit-learn-general] SPAMS python binding

> Is SPAMS Lasso a LARS or a Coordinate Descent implementation? LARS but I think there is a CD somewhere too Alex -- Better than sec? Nothing is better than sec when it comes to monitoring Big Data applications. Try Boun

Re: [Scikit-learn-general] SPAMS python binding

Le 18 avril 2012 05:33, Alexandre Gramfort a écrit : > hi sklearners, > > FYI the SPAMS python bindings are now online. > > https://gforge.inria.fr/projects/spams-devel/ > > A quick bench script: > > https://gist.github.com/2413309 > > For the lazy ones: > > http://twitpic.com/9b8bmk > > Take home

[Scikit-learn-general] SPAMS python binding

hi sklearners, FYI the SPAMS python bindings are now online. https://gforge.inria.fr/projects/spams-devel/ A quick bench script: https://gist.github.com/2413309 For the lazy ones: http://twitpic.com/9b8bmk Take home message the LARS in scikit-learn is just slightly slower. do not look too m