You can use the following basic syntax to get the top N rows by group in a pandas DataFrame:

df.groupby('group_column').head(2).reset_index(drop=True)

This particular syntax will return the top 2 rows by group.

Simply change the value inside the head() function to return a different number of top rows.

The following examples show how to use this syntax with the following pandas DataFrame:

import pandas as pd

#create DataFrame
df = pd.DataFrame({'team': ['A', 'A', 'A', 'A', 'A', 'B', 'B', 'B', 'B', 'B'],
                   'position': ['G', 'G', 'G', 'F', 'F', 'G', 'G', 'F', 'F', 'F'],
                   'points': [5, 7, 7, 9, 12, 9, 9, 4, 7, 7]})

#view DataFrame
print(df)

  team position  points
0    A        G       5
1    A        G       7
2    A        G       7
3    A        F       9
4    A        F      12
5    B        G       9
6    B        G       9
7    B        F       4
8    B        F       7
9    B        F       7

Example 1: Get Top N Rows Grouped by One Column

The following code shows how to return the top 2 rows, grouped by the team variable:

#get top 2 rows grouped by team
df.groupby('team').head(2).reset_index(drop=True)

        team	position  points
0	A	G	  5
1	A	G	  7
2	B	G	  9
3	B	G	  9

The output displays the top 2 rows, grouped by the team variable.

Example 2: Get Top N Rows Grouped by Multiple Columns

The following code shows how to return the top 2 rows, grouped by the team and position variables:

#get top 2 rows grouped by team and position
df.groupby(['team', 'position']).head(2).reset_index(drop=True)
	team	position  points
0	A	G	  5
1	A	G	  7
2	A	F	  9
3	A	F	  12
4	B	G	  9
5	B	G	  9
6	B	F	  4
7	B	F	  7

The output displays the top 2 rows, grouped by the team and position variables.

The following tutorials explain how to perform other common operations in pandas: