Filling nan values if unique conditions on multiple columns are met in Pandas

Question

I want to fill col3 nan values if the combination of col1 and col2 are matched in another row (but only once). For example: on row 2, since col1 and col2 are both cat, then fill col3 with 'George'. On row 4, since col1 and col2 are Cat and Dog, then fill col3 with Matt

However, if the combination has occurred more than once like in rows 5 and 7, then ignore don't fill nan with anything.

col1	col2	col3
Cat	Cat	George
Cat	Cat	nan
Cat	Dog	Matt
Cat	Dog	nan
Parrot	Elk	John
Parrot	Elk	nan
Parrot	Elk	Ryan
Parrot	Elk	nan

col1	col2	col3
Cat	Cat	George
Cat	Cat	George
Cat	Dog	Matt
Cat	Dog	Matt
Parrot	Elk	John
Parrot	Elk	nan
Parrot	Elk	Ryan
Parrot	Elk	nan

I hope this makes sense. Thanks

Answer 1

We can first get the sizes of each group col1, col3. Then those of size two we use ffill on:

sizes = df.groupby(['col1', 'col2'])['col2'].transform("size")
df.loc[sizes <= 2, "col3"] = df.loc[sizes <= 2, "col3"].ffill()

     col1 col2    col3
0     Cat  Cat  George
1     Cat  Cat  George
2     Cat  Dog    Matt
3     Cat  Dog    Matt
4  Parrot  Elk    John
5  Parrot  Elk     NaN
6  Parrot  Elk    Ryan
7  Parrot  Elk     NaN