Groupby pandas dataframes

June 9, 2022

I have a stupid problem on pandas, and today I don’t find solution, I feel it’s simple but a little bit complicated to explain…
Quickly, I have this dataframe :

	id_client	name_client	purchases
0	26	John	Alpha
1	26	John	Beta
2	26	John	Omega
3	45	Michael	Alpha
4	45	Michael	Beta
5	45	Michael	Omega
6	78	Charlie	Alpha
7	78	Charlie	Beta

code to reproduce :

df = pd.DataFrame({'id_client':[26,26,26,45,45,45,78,78], 'name_client':["John","John","John","Michael","Michael","Michael","Charlie","Charlie"], 'purchases':["Alpha","Beta","Omega","Alpha","Beta","Omega","Alpha","Beta"]})

And I want this result :

	id_client	name_client	options
0	26	John	Alpha, Beta, Omega
1	45	Michael	Alpha, Beta, Omega
2	78	Charlie	Alpha, Beta

I try :

df.groupby(by=['id_client']).transform(lambda x : ', '.join(x))

But I Have this bad result :

	name_client	purchases
0	John, John, John	Alpha, Beta, Omega
1	John, John, John	Alpha, Beta, Omega
2	John, John, John	Alpha, Beta, Omega
3	Michael, Michael, Michael	Alpha, Beta, Omega
4	Michael, Michael, Michael	Alpha, Beta, Omega
5	Michael, Michael, Michael	Alpha, Beta, Omega
6	Charlie, Charlie	Alpha, Beta
7	Charlie, Charlie	Alpha, Beta

I tried some differents options with groupby() but I don’t find the right solution
Thanks you.

>Solution :

Try:

>>> df.groupby(["id_client","name_client"], as_index=False).agg(", ".join)

   id_client name_client           purchases
0         26        John  Alpha, Beta, Omega
1         45     Michael  Alpha, Beta, Omega
2         78     Charlie         Alpha, Beta