ENH: Inconsistency when string refers both to index level and column label

[ChrisStuff]

While groupby throws a ValueError when the string parameter refers both to an index level and column name, query gives the column precedence in such a case:

>>> df = pd.DataFrame({'a':[1,2,3,4,5], 'b':[3,3,3,3,3]})
>>> df.index.name = 'a'
>>> df.query('a < b')
   a  b
a
0  1  3
1  2  3
>>> df.groupby('a') 
Traceback (most recent call last):
ValueError: 'a' is both an index level and a column label, which is ambiguous.

I am sure there are many other places where a string can refer to both an index level and column name. #27652 and #8162 are somehow linked.

Even though those issues might not be solved, it would be desirable to handle the above ambiguity consistently in places where both index levels and column labels can be referenced by a string argument.

[jreback]

please an example

[TomAugspurger]

I think this one is not ambiguous since df.query('a < b') is equivalent to df['a'] < df['b'] and there's no ambiguity about what df['a'] refers to.

The groupby case is ambiguous since by='a' can be either the column or index level.