Using a custom function in chained polars code

pipe

Some functions you write in your code may be too specific to exist as built-in Polars methods or expressions. Fortunately, Polars provides a way to apply your custom function to a dataframe within a chained operation. Below is a dataframe of Apple stock data.

shape: (10, 2)

Date	Open
date	f64
2024-01-02	187.149994
2024-01-03	184.220001
2024-01-04	182.149994
2024-01-05	181.990005
2024-01-08	182.089996
2024-01-09	183.919998
2024-01-10	184.350006
2024-01-11	186.539993
2024-01-12	186.059998
2024-01-16	182.160004

Create custom function

Suppose you have the following custom function to calculate the change in value between the previous day’s Open and the current day’s Open.

def delta_change(df, column):
    return df.with_columns(
        (pl.col(column) - pl.col(column).shift(-1)).alias('Delta Change')
    )

This function takes two arguments: the dataframe and the Open column.

Use custom function

To use a custom function in a chained operation in Polars, you can use the pipe method like this:

df.pipe(delta_change, column='Open')

shape: (10, 3)

Date	Open	Delta Change
date	f64	f64
2024-01-02	187.149994	2.929993
2024-01-03	184.220001	2.070007
2024-01-04	182.149994	0.159989
2024-01-05	181.990005	-0.099991
2024-01-08	182.089996	-1.830002
2024-01-09	183.919998	-0.430008
2024-01-10	184.350006	-2.189987
2024-01-11	186.539993	0.479995
2024-01-12	186.059998	3.899994
2024-01-16	182.160004	null

My Polars course has over 150 students. Enroll to be part of the happy learners.