Creating a datetime column from multiple columns in polars

pl.date

When you have a column with date values, having its data type as datetime is beneficial because it allows you to perform time series analysis on your data. You can slice the data into months, weeks, days, etc. This is not possible when your date values are stored as a string (text) data type.

Below is a dataframe containing three columns: Day, Month, and Year.

shape: (4, 3)

Day	Month	Year
i64	i64	i64
23	12	1997
14	9	2001
18	7	2003
15	10	1999

You now want to create a new column, Date, that will contain all the values from each column as a full date. Normally, to create this new column, you would have to convert the values in each column to strings and then concatenate them into a single date. Then, you would need to change the data type of that full date from text to datetime. Fortunately, there’s a better way.

The better way

Polars has a handy expression, pl.date, that allows you to concatenate these values into a new column with a datetime data type. Here’s how to do it:

(df
 .with_columns(pl.date(pl.col('Year'), pl.col('Month'), pl.col('Day')))
 )

shape: (4, 4)

Day	Month	Year	date
i64	i64	i64	date
23	12	1997	1997-12-23
14	9	2001	2001-09-14
18	7	2003	2003-07-18
15	10	1999	1999-10-15

Notice that the new column has date as its data type.

Join 100+ students in my Polars course sharpening their data analysis skills.