Closed
Description
One feature about DataFrame that I found uncomfortable is the inconsistency of results:
x = pd.DataFrame({'A': [1,2,3,4,5]}, index=['a','b','c','d','e'])
x.loc[['a', 'f']] # fine, value of 'f' is numpy.nan
x.loc[['f']] # throws
When developing data applications, I spend lot of time fixing this issue. I think we should make it more consistent. Either always returns (preferred) or always throws.