Is there a way in pandas to do groupby and count without naming a specific column?

Asked Sep 27 '19 at 09:37

Active Sep 27 '19 at 09:51

Viewed 41 times

Is there a way in Pandas to groupby() and count() without naming a specific column? So typically in a data frame (df) with Columns A-D I could do

df.groupby(["A","B"]).count()

which will give me a two rows (C,D) with the count of non-empty (non-Nan) values of C, D where A and B have the same value. That is all nice, but oftentimes I'm just interested how many rows there are with the same A and B combination independent of what C and D are called at the moment and what their values are.

I can also just pick one of the columns and just get one column with the counts

df.groupby(["A","B"])["c"].count()

But for that I need to ensure that C is always there and is named "C". Sure I could include a dummy column

df.assign(dummy=1).groupby(["A","B"])["dummy"].count()

but I'm wondering if there is not more strait forward way.

edited Sep 27 '19 at 09:51

asked Sep 27 '19 at 09:37

Magellan88

2,543
3
24
36

1

try: `df.groupby('A').size()` – Andy L. Sep 27 '19 at 09:44
1

`df['A'].value_counts()` – Aryerez Sep 27 '19 at 09:45
I see, very good points, however i need to actually aggregate on two columns... I'll adapt the question. – Magellan88 Sep 27 '19 at 09:49
can you create a small dataframe as example and add expected output too? – anky Sep 27 '19 at 09:53
does `df.groupby(['A', 'B']).size()` help? – henrywongkk Sep 27 '19 at 09:53

Is there a way in pandas to do groupby and count without naming a specific column?

0 Answers0