Scoring for dummy classifier does not work without test samples

[JarnoRFB]

When using the Dummy classifier, it is possible to fit the classifier without providing any examples, which makes sense as the classifier only operates on the targets. However, it is not possible to score the classifier without providing examples. Using DummyClassifier without examples is helpful if the examples, but not the targets, are to big to fit into memory. One can still get around this by constructing
artificial examples, say just zeros, but avoiding this would make things a bit easier.

Code to Reproduce

from sklearn.dummy import DummyClassifier
import numpy as np

y = [1, 1, 2]
y_ = [1, 1, 2]

d = DummyClassifier()
d.fit(None, y)
print(d.score(None, y_))

Expected Results

The score is printed.

Actual Results

An exception is thrown

ValueError: Found input variables with inconsistent numbers of samples: [3, 1]

So one has to do

from sklearn.dummy import DummyClassifier
import numpy as np

y = [1, 1, 2]
y_ = [1, 1, 2]
x = np.zeros(shape=(3, 1))

d = DummyClassifier()
d.fit(None, y)
print(d.score(x,  y_))

If this seems like a useful feature, I would be happy to submit a PR.

Versions

Linux-4.15.0-33-generic-x86_64-with-debian-buster-sid
Python 3.6.4 |Anaconda, Inc.| (default, Jan 16 2018, 18:10:19)
[GCC 7.2.0]
NumPy 1.14.3
SciPy 1.0.0
Scikit-Learn 0.19.2

[jnothman]

I suppose this would be consistent. PR welcome