BUG: iloc misbehavior with pd.Series: sometimes returns pd.Categorical instead

[navin-k80]

A small, complete example of the issue

# Your code here
>>> type(pd.Series([1,2,3]).astype('category').iloc[0:1])
<class 'pandas.core.series.Series'>

>>> type(pd.Series([1,2,3]).astype('category').iloc[np.array([0, 1])])
<class 'pandas.core.categorical.Categorical'>

Expected Output

Both should return a pandas Series object

Output of pd.show_versions()

pandas version = 0.18

[jreback]

in 0.19.0

In [18]: pd.Series([1,2,3]).astype('category').iloc[0:1]
Out[18]: 
0    1
dtype: category
Categories (3, int64): [1, 2, 3]

In [19]: pd.Series([1,2,3]).astype('category').iloc[np.array([0, 1])]
Out[19]: 
[1, 2]
Categories (3, int64): [1, 2, 3]

could be #12531 (in 0.18.1) or in 0.19.0

[jreback]

actually, I stand corrected. This is a different treatment of slices vs list-like
so this is a bug.

In [16]: s = pd.Series([1,2,3]).astype('category')

In [17]: type(s.iloc[[0,1]])
Out[17]: pandas.core.categorical.Categorical

In [18]: type(s.iloc[0:1])
Out[18]: pandas.core.series.Series