Python unique concatenate value of a column in another-CodePudding

I'm trying to perform this process, imagine to have the following and I want to obtain col4. :

Col1	Col2	Col3	Col 4	Col 5
SF	123	QW	QW, BF	1
SF	456	AF	AF	2
SO	xxx	AF	AF, BF	3
SO	yyy	GD	GD	4
SF	123	BF	QW, BF	1
RE	xxx	BF	AF, BF	5

For the purpose of aggragation I'm using these 2 lines of code:

df[df['col1']!='SF'].groupby(['Col2']).agg({'Col3' : lambda x: ','.join(x.unique())})

df[df['col1']=='SF'].groupby(['Col2','Col5']).agg({'Col3':','.join})

But I don't know how to put them on df. I tried also a merge but didn't work. I only hope to have been clear!!

Thanks so much in advance

EDIT 1 Sorry for not being clear. Before to perform any line of code I have Col1, Col2, Col3, Col5. Col4 is the output I would like to obtain.

CodePudding user response：

You transform instead of agg to assign back to the original DataFrame:

df["Col4"] = df.groupby("Col2")["Col3"].transform(lambda x: ", ".join(x.unique()))