pandas group by ALL functionality?

Question

I'm using the pandas groupby+agg functionality to generate nice reports

aggs_dict = {'a':['mean', 'std'], 'b': 'size'}
df.groupby('year').agg(aggs_dict)

I would like to use the same aggs_dict on the entire dataframe as a single group, with no division to years, something like:

df.groupall().agg(aggs_dict)

or:

df.agg(aggs_dict)

But couldn't find any elegant way to do it.. Note that in my real code aggs_dict is quite complex so it's rather cumbersome to do:

df.a.mean()
df.a.std()
df.b.size()
....

am I missing something simple and nice?

@ayhan IIUC, it's the opposite - if the entire index would be one big duplicate, that would work here. The question is about an aggregation for the entire df as a group, not for each of the rows. — Ami Tavory, Sep 07 '16 at 08:41

score 12 · Answer 1 · answered Oct 05 '17 at 10:30

12

You could also use a function to directly group on:

 df.groupby(lambda x: True).agg(aggs_dict)

answered Oct 05 '17 at 10:30

Hervé Mignot

131
1
2

score 7 · Accepted Answer · answered Sep 07 '16 at 12:45

7

Ami Tavory's answer is a great way to do it but just in case you wanted a solution that doesn't require creating new columns and deleting them afterwards you could do something like:

df.groupby([True]*len(df)).agg(aggs_dict)

answered Sep 07 '16 at 12:45

bunji

4,366
14
29

wow! exactly what I wanted (cumbersome notation, but I'm used to that from pandas :) – ihadanny Sep 07 '16 at 19:32
Nice answer! – Ami Tavory Sep 07 '16 at 20:42

score 5 · Answer 3 · edited May 23 '17 at 10:29

5

You could add a dummy column:

df['dummy'] = 1

Then groupby + agg on it:

df.groupby('dummy').agg(aggs_dict)

and then delete it when you're done.

edited May 23 '17 at 10:29

Community

1
1

answered Sep 07 '16 at 08:39

Ami Tavory

66,807
9
114
153

pandas group by ALL functionality?

3 Answers3