Pandas transform() vs apply()

Question

I don\'t understand why apply and transform return different dtypes when called on the same data frame. The way I explained the two functions to my

Answer 1

Just adding another illustrative example with sum as I find it more explicit:

df = (
    pd.DataFrame(pd.np.random.rand(10, 3), columns=['a', 'b', 'c'])
        .assign(a=lambda df: df.a > 0.5)
)

Out[70]: 
       a         b         c
0  False  0.126448  0.487302
1  False  0.615451  0.735246
2  False  0.314604  0.585689
3  False  0.442784  0.626908
4  False  0.706729  0.508398
5  False  0.847688  0.300392
6  False  0.596089  0.414652
7  False  0.039695  0.965996
8   True  0.489024  0.161974
9  False  0.928978  0.332414

df.groupby('a').apply(sum)  # drop rows

         a         b         c
a                             
False  0.0  4.618465  4.956997
True   1.0  0.489024  0.161974


df.groupby('a').transform(sum)  # keep dims

          b         c
0  4.618465  4.956997
1  4.618465  4.956997
2  4.618465  4.956997
3  4.618465  4.956997
4  4.618465  4.956997
5  4.618465  4.956997
6  4.618465  4.956997
7  4.618465  4.956997
8  0.489024  0.161974
9  4.618465  4.956997

However when applied to pd.DataFrame and not pd.GroupBy object I was not able to see any difference.