Rounding down time in polars

dt.truncate

Polars allows you to round down datetime values, similar to how you would round decimal values–well, almost. Below is a dataframe showing time series data.

shape: (5, 1)

Timestamp
datetime[μs]
2023-01-01 00:00:00
2023-01-01 02:05:00
2023-01-01 04:10:00
2023-01-01 06:15:00
2023-01-01 08:20:00

Round down time values

Here, we use Polars to scan a 10-minute interval of the time series and set each value to the lower bound of its interval. For example, minutes between 0 and 10 become 0, minutes between 10 and 20 become 10, and so on. Essentially, we are rounding down the values to the specified interval of 10 minutes.

(df
 .with_columns(pl.col('Timestamp').dt.truncate('10m')
               .alias('Truncated_by_10min'))
 )

shape: (5, 2)

Timestamp	Truncated_by_10min
datetime[μs]	datetime[μs]
2023-01-01 00:00:00	2023-01-01 00:00:00
2023-01-01 02:05:00	2023-01-01 02:00:00
2023-01-01 04:10:00	2023-01-01 04:10:00
2023-01-01 06:15:00	2023-01-01 06:10:00
2023-01-01 08:20:00	2023-01-01 08:20:00

Learn more in my Polars course.