How to do 'groupby' a column value (str) and return n unique rows in pandas

Question

I have this DataFrame

df1=
     col1 col2 col3 value
      A   B     AB   20
      C   C     AB   10
      D   C     AB   90
      E   E     AA   10
      F   E     AA   20
      G   F     AA   10

I want to groupby column 'col3' and return two rows for each col3 value(random).something like this.

    df1=
     col1 col2 col3 value
      A   B     AB   20

      D   C     AB   90

      F   E     AA   20

      G   F     AA   10

Use `df.sample(frac=1).groupby('col3').head(2)` – jezrael Apr 24 '20 at 12:14 — jezrael, Apr 24 '20 at 12:14

score 1 · Accepted Answer · answered Apr 24 '20 at 12:11

1

Try this:

df.groupby('col3').apply(lambda x: x.sample(2)).reset_index(drop=True)

OR with slightly better syntax:

df.groupby('col3').apply(pd.DataFrame.sample,n=2).reset_index(drop=True)

answered Apr 24 '20 at 12:11

Mayank Porwal

33,470
8
37
58

How to do 'groupby' a column value (str) and return n unique rows in pandas

1 Answers1