On Saturday, September 12, 2015, Andy <t3k...@gmail.com> wrote: > On 09/12/2015 02:54 AM, Daan Wynen wrote: > > Hi Andy, > > according to [1] "The multiclass support is handled according to a > one-vs-one scheme." > That's why I was using the wrapper. > > SVC has one-vs-rest built-in. >
actually, it has one-vs-one built-in, at least according to the docs and from what I recall from when using it. svc.dual_coef_ has one axis of shape n_classes -1, which is an indicator for this. That means that if you want OvR, you need to put it in the meta-estimator. > > What should the docs say? > > The multiclass docs here are quite explicit: > http://scikit-learn.org/stable/modules/multiclass.html > > Warning > > All classifiers in scikit-learn do multiclass classification > out-of-the-box. You don’t need to use the sklearn.multiclass > <http://scikit-learn.org/stable/modules/classes.html#module-sklearn.multiclass> > module unless you want to experiment with different multiclass strategies. >
------------------------------------------------------------------------------
_______________________________________________ Scikit-learn-general mailing list Scikit-learn-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
