BUG: Series.any/all with level keyword should preserve boolean dtype

[jorisvandenbossche]

In [5]: s = pd.Series([False, False, True, True, False, True],
                      index=[0, 0, 1, 1, 2, 2], dtype="boolean")  

In [6]: s.all(level=0) 
Out[6]: 
0    False
1     True
2    False
dtype: bool

The dtype of the result could also be boolean instead of bool.

[KenilMehta]

Hi @jorisvandenbossche
This is my first Pandas Issue and I would like to help solve this issue.

Just for understanding the issue, the dtype of the Series object is "boolean", hence, we want the return type of Series.all() also to be "boolean" instead of "bool". Is that right?

[jorisvandenbossche]

Yes, that is correct. But specifically with the level keyword as in the example above. Without using the level keyword, it already works as expected.

[KenilMehta]

Do we want the dtype of result Series to match with the original Series in all the cases or only in the case of boolean?

What would you expect the dtype of result Series to be in the following example?

import pandas as pd
import numpy as np

s = pd.Series([0, 0, 1, 1, 0, 1], index=[0, 0, 1, 1, 2, 2], dtype="int64")
print(s.all(level=0))