Table of Contents
Pandas Groupby is a powerful tool that can be used to apply a function to a group of data. To do this, you need to first define a function that you would like to apply. Then, use the groupby() method to group the data into the desired groups. Finally, apply the function to the grouped data using the apply() method. This will apply the function to each group and return the results in a DataFrame format.
You can use the following basic syntax to use the groupby() and apply() functions together in a pandas DataFrame:
df.groupby('var1').apply(lambda x: some function)
The following examples show how to use this syntax in practice with the following pandas DataFrame:
import pandas as pd #create DataFrame df = pd.DataFrame({'team': ['A', 'A', 'A', 'B', 'B', 'B', 'B'], 'points_for': [18, 22, 19, 14, 11, 20, 28], 'points_against': [14, 21, 19, 14, 12, 20, 21]}) #view DataFrame print(df) team points_for points_against 0 A 18 14 1 A 22 21 2 A 19 19 3 B 14 14 4 B 11 12 5 B 20 20 6 B 28 21
Example 1: Use groupby() and apply() to Find Relative Frequencies
The following code shows how to use the groupby() and apply() functions to find the relative frequencies of each team name in the pandas DataFrame:
#find relative frequency of each team name in DataFrame
df.groupby('team').apply(lambda x: x['team'].count() / df.shape[0])
team
A 0.428571
B 0.571429
dtype: float64
From the output we can see that team A occurs in 42.85% of all rows and team B occurs in 57.14% of all rows.
Example 2: Use groupby() and apply() to Find Max Values
The following code shows how to use the groupby() and apply() functions to find the max “points_for” values for each team:
#find max "points_for" values for each team
df.groupby('team').apply(lambda x: x['points_for'].max())
team
A 22
B 28
dtype: int64
From the output we can see that the max points scored by team A is 22 and the max points scored by team B is 28.
Example 3: Use groupby() and apply() to Perform Custom Calculation
The following code shows how to use the groupby() and apply() functions to find the mean difference between “points_for” and “points_against” for each team:
#find max "points_for" values for each team
df.groupby('team').apply(lambda x: (x['points_for'] - x['points_against']).mean())
team
A 1.666667
B 1.500000
dtype: float64
From the output we can see that the mean difference between “points for” and “points against” is 1.67 for team A and 1.50 for team B.