Pandas dataframe.groupby()

The groupby statement in DataFrame groups rows that have the similar values into summary rows, like "find the number of Apple Steve have". The DataFrame groupby statement is often used with aggregate functions (sum, count, mean, min, max etc.) to group the output by one or more columns.

Lets' create DataFrame with values.

 
import pandas as pd
import numpy as np
df = pd.DataFrame()
df['Name'] = ['Doe', 'Doe', 'Mike','Steve','Doe','Doe','Mike','Doe','Mike']
df['Fruit'] = ['Apple', 'Apple', 'Apple','Orange','Orange','Orange','Orange','Grapes','Grapes']
df['Count'] = [20, 10, 20,30,10,40,50,10,30]
df

 
    Name   Fruit  Count
0    Doe   Apple     20
1    Doe   Apple     10
2   Mike   Apple     20
3  Steve  Orange     30
4    Doe  Orange     10
5    Doe  Orange     40
6   Mike  Orange     50
7    Doe  Grapes     10
8   Mike  Grapes     30

Here you can see 3 names (Doe, Mike and Steve) have different kind of fruits (Apple, Orange and Grapes). So, you can have some operations on these tables using DataFrame groupby statement.

In the above image you can see some results from the above DataFrame. So, lets try to get the above result using DataFrame group by operation.

 
df.groupby(['Name','Fruit']).sum()

 
              Count
Name  Fruit
Doe   Apple      30
      Grapes     10
      Orange     50
Mike  Apple      20
      Grapes     30
      Orange     50
Steve Orange     30

Apply reset_index()

 
df.groupby(['Name','Fruit'])['Count'].sum().reset_index()

 
    Name   Fruit  Count
0    Doe   Apple     30
1    Doe  Grapes     10
2    Doe  Orange     50
3   Mike   Apple     20
4   Mike  Grapes     30
5   Mike  Orange     50
6  Steve  Orange     30

Also, you get another result to change the groupby order:

 
df.groupby(['Fruit','Name']).sum()

 
              Count
Fruit  Name
Apple  Doe       30
       Mike      20
Grapes Doe       10
       Mike      30
Orange Doe       50
       Mike      50
       Steve     30

Pivot Table

You can use the pivot functionality to arrange the data in a better grid.

 
df.groupby(['Name','Fruit'],as_index = False).sum().pivot('Name','Fruit').fillna(0)

 
Fruit Apple Grapes Orange
Name
Doe    30.0   10.0   50.0
Mike   20.0   30.0   50.0
Steve   0.0    0.0   30.0

Find the total count of fruits by person

 
df.groupby(['Name']).sum()

 
       Count
Name
Doe       90
Mike     100
Steve     30

Find the total count of fruits

 
df.groupby(['Fruit']).sum()

 
        Count
Fruit
Apple      50
Grapes     40
Orange    130

How many row entries for fruits in the table?

 
df.groupby(['Fruit']).size().reset_index()

 
    Fruit  0
0   Apple  3
1  Grapes  2
2  Orange  4

Next : Pandas DataFrame Aggregation and Grouping