Calculate moving average in polars

rolling_mean

You’re running a store and want to know the average of your sales every 3 days to track whether your sales are dipping. Below is a dataframe showing sales on specific weekdays.

3-day moving average

You can calculate the 3-day moving average in Polars using the expression rolling_mean like this:

(df
 .with_columns(Rol_Mean=pl.col('Sales').rolling_mean(window_size=3))
 )

shape: (8, 3)

Day	Sales	Rol_Mean
str	i64	f64
"Mon"	93	null
"Tue"	88	null
"Wed"	73	84.666667
"Thur"	69	76.666667
"Fri"	55	65.666667
"Sat"	89	71.0
"Sun"	92	78.666667
"Mon"	81	87.333333

Notice that the first 2 rows in Rol_Mean are null. That’s because the window size of 3 is not met on the first and second roll. If you don’t want null values, you can use the parameter min_samples, which determines the number of values that should be non-null before calculating the average. Let’s set it to 1.

(df
 .with_columns(Rol_Mean=pl.col('Sales').rolling_mean(window_size=3, min_samples=1))
 )

shape: (8, 3)

Day	Sales	Rol_Mean
str	i64	f64
"Mon"	93	93.0
"Tue"	88	90.5
"Wed"	73	84.666667
"Thur"	69	76.666667
"Fri"	55	65.666667
"Sat"	89	71.0
"Sun"	92	78.666667
"Mon"	81	87.333333

Enroll in my Polars course to advance your data analysis skills.