How to convert datetime64[ns] effective to other timezones?

[Rumpen]

Hi,

we have a basic Dataframe like this:

The process time is in UTC and we want to group by hour and aggregate by total_net using mean:

df['hour'] = df.process_time.dt.hour
group_by = df.groupby(by=df['hour'], sort="hour asc", agg={'aggregation': vaex.agg.mean(df['total_net'])})

Result would be:

Are there any best practices for converting the UTC timestamps to other Timezones using native vaex (like pandas .tz_localize and .tz_convert), e.g. User wants to see the aggregation in Timezone "Europe/Berlin" instead of "UTC"?

[JovanVeljanoski]

No native way i am afraid. I guess your best bet is to either do that prior to loading the data with vaex, or use some function (perhaps that of pandas) with apply. I vaex apply works in parallel, but best to be avoided if not necessary.

If this is a one-time thing conversion, you can "pay the price" with apply, and then maybe df.materialize the new column or store it to disk.

[Rumpen]

Thanks for the quick response and the good work on vaex!