[pymvpa] Can't access an enabled classifier attribute.

[Roberto Guidotti]

Thank you for the response Yaroslav,

Well the snippet probabilities = clf.ca.probabilities is only an example
that I've build to reveal the exception (I didn't think is a version
problem, I had 2.0.1 mvpa)!!
BTW I was quite sure that the problem relied on the cross validation :)!
 I would like to gather all the predictions/probabilities for each cv fold.
This why using the ca 'training_stats' during the definition of the
classifier, I had the probabilities of the samples during training?

So I can't have access to every cv fold probabilities/predictions using
CrossValidation object? Or I need a manual CrossValidation (and sometimes a
Feature Selection)?


Thank you!
Roberto

On 27 November 2012 15:49, Yaroslav Halchenko <debian at onerussian.com> wrote:

>
> On Tue, 27 Nov 2012, Roberto Guidotti wrote:
>
> >    Dear all,
> >    I've a problem with classifier's conditional attributes (it is
> possibly
> >    due to a bad command understanding, by myself, I admit!):
> >    clf = LinearCSVMC(C=1, probability=1, enable_ca=['probabilities'])
> >    cvte = CrossValidation(clf, NFoldPartitioner(cvtype = 1),
> >    enable_ca=['stats', 'repetition_results'])
> >    train_err = cvte(evds)
> >    When I try to see inside the conditional attributes of my trained
> >    classifier clf with the command below
> >    probabilities = clf.ca.probabilities
>
> odd enough above snippet (with probabilities = clf.ca.probabilities)
> works for me just fine.  but -- this way you are obtaining
> probabilities only for the last cross-validation fold.  what is the goal
> -- to get them for every CV fold? (then CrossValidation should get a
> callback to harvest them) or on all the data (then you would need to
> "clf.train(evds); clf.predict(evds);" and then gather
> clf.ca.probabilities
>
> >    I have this exception
> >    UnknownStateError: Unknown yet value of probabilities
> >    Instead using a classifier defined as:
> >    clf = LinearCSVMC(C=1, probability=1, enable_ca=['probabilities',
> >    'training_stats'])
> >    I could store my probabilities.
> >    Am I doing all the things well?
> >    Thank you
> >    Roberto
> --
> Yaroslav O. Halchenko
> Postdoctoral Fellow,   Department of Psychological and Brain Sciences
> Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
> Phone: +1 (603) 646-9834                       Fax: +1 (603) 646-1419
> WWW:   http://www.linkedin.com/in/yarik
>
> _______________________________________________
> Pkg-ExpPsy-PyMVPA mailing list
> Pkg-ExpPsy-PyMVPA at lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20121127/2de28963/attachment.html>