BUG: coercing of bools in groupby transform

[jongmmm]

Code Sample

df = pd.DataFrame({'A': [1.1,2.2], 'B':[1,2]})
df.groupby('B').A.transform(lambda x: True)

This returns

Out[3]:
0    1.0
1    1.0
Name: A, dtype: float64

Problem

groupby.transform returns numeric values, not bool . This is inconsistent
because DataFrame.transform behaves differently:

In [5]: df.A.transform(lambda x:True)
Out[5]:
0    True
1    True
Name: A, dtype: bool

Additional

But please note that if the original data type is int, it works properly:

In [9]: df = pd.DataFrame({'A': [1,2], 'B':[1,2]})
   ...: df.groupby('B').A.transform(lambda x: True)
   ...:
Out[9]:
0    True
1    True
Name: A, dtype: bool

[chris-b1]

Looks essentially the same as #10972, which was closed by #15430, but apparently didn't capture this case. apply does do the right inference here:

In [2]: df.groupby('B').A.apply(lambda x: True)
Out[2]:
B
1    True
2    True
Name: A, dtype: bool