I'd like to cast a column with the type datetype[ms]
to datetime[ns]
in polars. Is there a easy way to do this?
I think there's a problem with group_by_dynamic
when using datetime[ms]
and I'd like to test this
You can use .dt.cast_time_unit()
s = pl.datetime_range(
pl.datetime(2021, 1, 1), pl.datetime(2021, 1, 2),
interval="3h", time_unit="ns", eager=True
)
s.dt.cast_time_unit("ms")
shape: (9,)
Series: '' [datetime[ms]]
[
2021-01-01 00:00:00
2021-01-01 03:00:00
2021-01-01 06:00:00
2021-01-01 09:00:00
2021-01-01 12:00:00
2021-01-01 15:00:00
2021-01-01 18:00:00
2021-01-01 21:00:00
2021-01-02 00:00:00
]