max_by

Returns the value from the col parameter that is associated with the maximum value from the ord parameter. This function is often used to find the col parameter value corresponding to the maximum ord parameter value within each group when used with groupBy(). The function is non-deterministic so the output order can be different for those associated the same values of col.

Syntax

Python

from pyspark.sql import functions as sf

sf.max_by(col, ord)

Parameters

Parameter	Type	Description
col	pyspark.sql.Column or column name	The column representing the values to be returned. This could be the column instance or the column name as string.
ord	pyspark.sql.Column or column name	The column that needs to be maximized. This could be the column instance or the column name as string.

Returns

pyspark.sql.Column: A column object representing the value from col that is associated with the maximum value from ord.

Examples

Example 1: Using max_by with groupBy

Python

import pyspark.sql.functions as sf
df = spark.createDataFrame([
    ("Java", 2012, 20000), ("dotNET", 2012, 5000),
    ("dotNET", 2013, 48000), ("Java", 2013, 30000)],
    schema=("course", "year", "earnings"))
df.groupby("course").agg(sf.max_by("year", "earnings")).sort("course").show()

Output

+------+----------------------+
|course|max_by(year, earnings)|
+------+----------------------+
|  Java|                  2013|
|dotNET|                  2013|
+------+----------------------+

Example 2: Using max_by with different data types

Python

import pyspark.sql.functions as sf
df = spark.createDataFrame([
    ("Marketing", "Anna", 4), ("IT", "Bob", 2),
    ("IT", "Charlie", 3), ("Marketing", "David", 1)],
    schema=("department", "name", "years_in_dept"))
df.groupby("department").agg(
    sf.max_by("name", "years_in_dept")
).sort("department").show()

Output

+----------+---------------------------+
|department|max_by(name, years_in_dept)|
+----------+---------------------------+
|        IT|                    Charlie|
| Marketing|                       Anna|
+----------+---------------------------+

Example 3: Using max_by where ord has multiple maximum values

Python

import pyspark.sql.functions as sf
df = spark.createDataFrame([
    ("Consult", "Eva", 6), ("Finance", "Frank", 5),
    ("Finance", "George", 9), ("Consult", "Henry", 7)],
    schema=("department", "name", "years_in_dept"))
df.groupby("department").agg(
    sf.max_by("name", "years_in_dept")
).sort("department").show()

Output

+----------+---------------------------+
|department|max_by(name, years_in_dept)|
+----------+---------------------------+
|   Consult|                      Henry|
|   Finance|                     George|
+----------+---------------------------+