Issue with date time arguments

Feature Store
#1

Hello,

I have an issue with arguments that take datetime values.

My dataset has a feature of datetime for every hour with a datatype of period[H].

When I use the function FeatureView.get_batch_data and provide an argument to have the dataset start from 00:00 hours, it provides me the dataset from 12:00 hours. Similarly, with the end_time rather than providing the last datetime at 23:00 hours, it provides me with the last datetime of 11:00.

For eg. if I provide a datetime value of datetime.datetime(2023, 1, 1) then the dataset starts from ā€œ2023-01-01 12:00:00ā€.

I have tried multiple ways to get the dataset from the starting datetime of ā€œ2023-01-01 00:00:00ā€, but the only way is to subtract with datetime.timedelta.

#2

Hi @John_Pinto thanks for reporting this. This is issue from hopsworks side and it was addressed already. Upgrade will be released as soon as possible.

Meanwhile, if you indicate hours in datetime object it will give you more accurate time window

#3

Hi @John_Pinto - please let us know if we can help you with anything else. We are here to help - I have connected on LinkedIn as well for convenience.

#4

Hi @Davit_Bzhalava and @Rik_Van_Bruggen thanks for reaching out.

Currently, Iā€™m using Python 3.11.7 and Hopsworks 3.4.4. I have tried mentioning hours, minutes, and seconds, different combinations of these three, but still, the result is the same, the time windows are inaccurate.

To be more precise, if the input arguments are within the 12-hour period the output dataframe is accurate. If it is beyond 12 hours, for eg. The argument hour in datetime is either 0 or between 13-23 the output dataframe time window is 12 or between 1-11 hours with no changes to the date.

#5

Hi @John_Pinto, issue with serverless offering at the moment is that it has 12 hour clock. As I mentioned above it has been resolved in managed service where 24 hour clock was set. It will take some time until this fix reaches app.hopsworks.ai

Best regards,
Davit

#6

Hello @John_Pinto ,

The issue has been fixed on Serverless, so you can try again and let us know if you encounter any further issues.

Regards,
Mahmoud

1 Like