Looks good to me. +1

Can you confirm @rasbt ?

Certainly doesn't hurt in case someone uses float arrays to store class labels. That's actually a good idea, thanks for the fix!

I remember the "type in -- type out" discussion ;)

lr = LogisticRegression() 
>>> lr.fit(np.array([[1.0], [2.9]]), [1, 2]).predict([[1.0], [2.9]]).dtype
dtype('int64')

>>> lr.fit(np.array([[1.0], [2.9]]), [1.0, 2.0]).predict([[1.0], [2.9]]).dtype
dtype('float64')

I remember the "type in -- type out" discussion ;)

True, except that the classifiers are internally trained on the LabelEncoded outputs, so it should still spit the same format, even with this fix.

Thanks for the fix @alvarouc. I am merging since @rasbt seems OK with this as well.

no regression test? :-/

@amueller you mean that we should raise an exception if someone tries to seed it with an estimator for regression?

do you have full code example how it works

On Sun, Jan 24, 2016 at 1:37 PM, Sebastian Raschka <notifications@github.com

wrote:

@amueller https://github.com/amueller you mean that we should raise an
exception if someone tries to seed it with an estimator for regression?

—
Reply to this email directly or view it on GitHub
#5802 (comment)
.

@Sandy4321

do you have full code example how it works

What do you mean, exactly? A code example for raising the exception?

code example for Voting Classifier

On Sun, Jan 24, 2016 at 1:50 PM, Sebastian Raschka <notifications@github.com

wrote:

@Sandy4321 https://github.com/Sandy4321

do you have full code example how it works

What do you mean, exactly? A code example for raising the exception?

—
Reply to this email directly or view it on GitHub
#5802 (comment)
.

@Sandy4321 there are plenty of examples in the documentation at http://scikit-learn.org/stable/modules/ensemble.html#voting-classifier (or also see the respective unit tests)

(As well as the regression test), I think this needs a what's new entry.