BUG: factorize_from_iterables applies type to empty iterables

[drudd]

Code Sample, a copy-pastable example if possible

In [1]: import pandas as pd

In [2]: A = pd.MultiIndex(levels=[[], []], labels=[[], []], names=['a', 'b'])

In [3]: B = pd.MultiIndex.from_arrays(arrays=[[], []], names=['a', 'b'])

In [4]: A
Out[4]:
MultiIndex(levels=[[], []],
           labels=[[], []],
           names=[u'a', u'b'])

In [5]: B
Out[5]:
MultiIndex(levels=[[], []],
           labels=[[], []],
           names=[u'a', u'b'])

In [6]: pd.testing.assert_index_equal(A, B)
...
AssertionError: MultiIndex level [0] are different

MultiIndex level [0] classes are not equivalent
[left]:  Index([], dtype='object', name=u'a')
[right]: Float64Index([], dtype='float64', name=u'a')

Problem description

Empty iterables should be factorized to object types rather than Float64Index. This causes MultiIndexes created from the constructor to not equal those created using .from_arrays, which relies on factorization (ref PR #16782).

[mroeschke]

This looks to work on master. Could use a test.

In [276]: pd.__version__
Out[276]: '0.26.0.dev0+593.g9d45934af'

In [277]: In [1]: import pandas as pd
     ...:
     ...: In [2]: A = pd.MultiIndex(levels=[[], []], labels=[[], []], names=['a', 'b'])
     ...:
     ...: In [3]: B = pd.MultiIndex.from_arrays(arrays=[[], []], names=['a', 'b'])
/anaconda3/envs/pandas-dev/bin/ipython:3: FutureWarning: the 'labels' keyword is deprecated, use 'codes' instead
  # -*- coding: utf-8 -*-

In [278]: In [6]: pd.testing.assert_index_equal(A, B)
     ...:

In [279]: A
Out[279]: MultiIndex([], names=['a', 'b'])

In [280]: B
Out[280]: MultiIndex([], names=['a', 'b'])