Pandas groupby() Explained With Examples

Similar to the SQL GROUP BY clause pandas DataFrame.groupby() function is used to collect identical data into groups and perform aggregate functions on the grouped data. Group by operation involves splitting the data, applying some functions, and finally aggregating the results.

In pandas, you can use groupby() with the combination of sum(), pivot(), transform(), aggregate() and many more methods. In this article, I will cover how to group by a single column, or multiple columns by using aggregations with examples.

1. Pandas groupby() Syntax

Below is the syntax of the groupby() function, this function takes several params that are explained below and returns DataFrameGroupBy object that contains information about the groups.

# Syntax of DataFrame.groupby()
DataFrame.groupby(by=None, axis=0, level=None, as_index=True, 
       sort=True, group_keys=True, squeeze=<no_default>, 
       observed=False, dropna=True)

• by – List of column names to group by
• axis – Default to 0. It takes 0 or ‘index’, 1 or ‘columns’
• level – Used with MultiIndex.
• as_index – sql style grouped otput.
• sort – Default to True. Specify whether to sort after group
• group_keys – add group keys or not
• squeeze – depricated in new versions
• observed – This only applies if any of the groupers are Categoricals.
• dropna – Default to False. Use True to drop None/Nan on sory keys

In order to explain several examples of how to perform group by, first, let’s create a simple DataFrame with the combination of string and numeric columns.

import pandas as pd
technologies   = ({
    'Courses':["Spark","PySpark","Hadoop","Python","Pandas","Hadoop","Spark","Python","NA"],
    'Fee' :[22000,25000,23000,24000,26000,25000,25000,22000,1500],
    'Duration':['30days','50days','55days','40days','60days','35days','30days','50days','40days'],
    'Discount':[1000,2300,1000,1200,2500,None,1400,1600,0]
          })
df = pd.DataFrame(technologies)
print(df)

Yields below output.

   Courses    Fee Duration  Discount
0    Spark  22000   30days    1000.0
1  PySpark  25000   50days    2300.0
2   Hadoop  23000   55days    1000.0
3   Python  24000   40days    1200.0
4   Pandas  26000   60days    2500.0
5   Hadoop  25000   35days       NaN
6    Spark  25000   30days    1400.0
7   Python  22000   50days    1600.0
8       NA   1500   40days       0.0

2. Pandas groupby() Example

As I said above groupby() function returns DataFrameGroupBy object after collecting the identical data into groups from pandas DataFrame. This object contains several methods (sum(), mean() e.t.c) that can be used to aggregate the grouped rows.

# Use groupby() to compute the sum
df2 =df.groupby(['Courses']).sum()
print(df2)

Yields below output.

           Fee  Discount
Courses                 
Hadoop   48000    1000.0
NA        1500       0.0
Pandas   26000    2500.0
PySpark  25000    2300.0
Python   46000    2800.0
Spark    47000    2400.0

3. Pandas groupby() on Two or More Columns

Most of the time we would need to perform groupby on multiple columns of DataFrame, you can do this by passing a list of column labels you wanted to perform group by on.

# Group by multiple columns
df2 =df.groupby(['Courses', 'Duration']).sum()
print(df2)

Yields below output

                    Fee  Discount
Courses Duration                 
Hadoop  35days    25000       0.0
        55days    23000    1000.0
NA      40days     1500       0.0
Pandas  60days    26000    2500.0
PySpark 50days    25000    2300.0
Python  40days    24000    1200.0
        50days    22000    1600.0
Spark   30days    47000    2400.0

4. Add Index to the grouped data

By default groupby() result doesn’t include row Index, you can add the index using DataFrame.reset_index() method.

# Add Row Index to the group by result
df2 = df.groupby(['Courses','Duration']).sum().reset_index()
print(df2)

Yields below output

   Courses Duration    Fee  Discount
0   Hadoop   35days  25000       0.0
1   Hadoop   55days  23000    1000.0
2       NA   40days   1500       0.0
3   Pandas   60days  26000    2500.0
4  PySpark   50days  25000    2300.0
5   Python   40days  24000    1200.0
6   Python   50days  22000    1600.0
7    Spark   30days  47000    2400.0

5. Drop NA /None/Nan (on group key) from Result

You can also choose whether to include NA/None/Nan in group keys or not by setting dropna parameter. By default the value of dropna set to True. so to not to include None/Nan values on group keys set dropna=False parameter.

# Drop rows that have None/Nan on group keys
df2=df.groupby(by=['Courses'], dropna=False).sum()
print(df2)

6. Sort groupby() result by Group Key

By default groupby() function sorts results by group key hence it will take additional time, if you have a performance issue and don’t want to sort the group by the result, you can turn this off by using the sort=False param.

# Remove sorting on grouped results
df2=df.groupby(by=['Courses'], sort=False).sum()
print(df2)

If you wanted to sort key descending order, use below.

# Sorting group keys on descending order
groupedDF = df.groupby('Courses',sort=False).sum()
sortedDF=groupedDF.sort_values('Courses', ascending=False)
print(sortedDF)

In case you wanted to sort by a different key, you can do so by using DataFrame.apply() function.

# Using apply() & lambda
df.groupby('Courses').apply(lambda x: x.sort_values('Fee'))

7. Apply More Aggregations

You can also compute several aggregations at the same time in pandas by passing the list of agg functions to the aggregate().

# Groupby & multiple aggregations
result = df.groupby('Courses')['Fee'].aggregate(['min','max'])
print(result)

Yields below output.

           min    max
Courses              
Hadoop   23000  25000
NA        1500   1500
Pandas   26000  26000
PySpark  25000  25000
Python   22000  24000
Spark    22000  25000

The above example calculates min and max on the Fee column. Let’s extend this to compute different aggregations on different columns.

Note that applying several aggregations to a single column in pandas DataFrame will result in a MultiIndex.

# Groupby multiple columns & multiple aggregations
result = df.groupby('Courses').aggregate({'Duration':'count','Fee':['min','max']})
print(result)

Yields below output. Notice that this creates MultiIndex. Working with multi-indexed columns is not easy so I’d recommend flattening by renaming the columns.

        Duration    Fee       
           count    min    max
Courses                       
Hadoop         2  23000  25000
NA             1   1500   1500
Pandas         1  26000  26000
PySpark        1  25000  25000
Python         2  22000  24000
Spark          2  22000  25000

8. Complete Example of groupby() in DataFrame

import pandas as pd
technologies   = ({
    'Courses':["Spark","PySpark","Hadoop","Python","Pandas","Hadoop","Spark","Python","NA"],
    'Fee' :[22000,25000,23000,24000,26000,25000,25000,22000,1500],
    'Duration':['30days','50days','55days','40days','60days','35days','30days','50days','40days'],
    'Discount':[1000,2300,1000,1200,2500,None,1400,1600,0]
          })
df = pd.DataFrame(technologies)
print(df))

# Use groupby() to compute the sum
df2 =df.groupby(['Courses']).sum()
print(df2)

# Group by on multiple columns
df2 =df.groupby(['Courses', 'Duration']).sum()
print(df2)

# Set Index on group by results
df2 = df.groupby(['Courses','Duration']).sum().reset_index()
print(df2)

# Using groupby on single column in pandas 
df2 = df.groupby(['Courses'])['Fee'].sum().reset_index()
print(df2)

# Ignore sorting on group by key
df2=df.groupby(by=['Courses'], sort=False).sum()
print(df2)

# Sort group key on descending order
groupedDF = df.groupby('Courses',sort=False).sum()
sortedDF=groupedDF.sort_values('Courses', ascending=False)
print(sortedDF)

# Groupby & multiple aggregations
result = df.groupby('Courses')['Fee'].aggregate(['min','max'])
print(result)

# Groupby multiple columns & multiple aggregations
result = df.groupby('Courses').aggregate({'Duration':'count','Fee':['min','max']})
print(result)

Conclusion

In this article, I have covered Pandas groupby() syntax and several examples of how to collect identical data into groups. I hope you have learned how to run group by on several columns, sort grouped data, ignore null values, and many more with examples.

Happy Learning !!

Related Articles

• Pandas DataFrame – Different Ways to Iterate Over Rows
• Different Ways to Get Row Count in Pandas DataFrame
• How to Shuffle Pandas Rows Randomly
• Pandas GroupBy Multiple Columns Explained
• How to GroupBy Index in Pandas?
• PySpark GroupBy Count – Explained
• Pandas Groupby Sort within Groups
• Pandas groupby() and count() with Examples