pandas groupby with agg not working on multiple columns

Question

I'm trying to merge multiple columns, each into a list based on a group by in pandas. Below is the code I'm using

grouped_df = df.groupby(['d_id', 'time']).agg({'d_name': lambda x: tuple(x)},  
{'ver': lambda x: tuple(x)},
{'f_name': lambda x: tuple(x)})

This only gives me the first column (d_name) in a list with d_id and time in grouped_df. The other two columns do not show as lists. I tried using list earlier but found out that list has an issue with agg function so I resorted to tuple. Let me know if I'm doing something wrong here.

Thanks to RafaelC for the answer to this. Now I am trying to add these list columns to the original dataframe as grouped_df. When I see the columns in grouped_df they come out as

Index([u'ver', u'f_name', u'd_name'], dtype='object')

But when I do a head, I get

 ver  \
d_id  time
1    2018-06-01   (ver1, ver2, ver3.....
2    2018-06-01   (ver1, ver2, ver3.....
3    2018-06-01   (ver1, ver2, ver3.....


f_name  \
d_id   time
1  2018-06-01   (f_name1, f_name2, f_name2.....
2  2018-06-01   (f_name1, f_name2, f_name2.....
3  2018-06-01   (f_name1, f_name2, f_name2.....

d_name
d_id   time
1 2018-06-01   (d_name1, dname2, d_name3...
2 2018-06-01   (d_name1, dname2, d_name3...
3 2018-06-01   (d_name1, dname2, d_name3...

How do I get the following d_id time ver d_name f_name where ver, d_name and f_name are lists.

Hope this is clear.

Borne

Thanks. Solves it like a charm :). If you could post it, I'll accept as an answer. – Borne

One additional thing though, the initial df columns are not part of the resulting dataframe ('d_id', 'time'). Any ideas on how to join the list columns back with the groupby columns? – Borne

hm, what do you mean? Can you edit the question with more details? – Ducan

Edited my question to add further details – Borne

Answer 1

Use a single argument instead of three

{'d_name': lambda x: tuple(x), 'ver': lambda x: tuple(x), 'f_name': lambda x: tuple(x)}

Ducan