Sub-daily holidays/events
- Dominant language
- Python
- Stars
- 20.4k
- Forks
- 4.6k
- Avg merge
- 19h 52m
- Merged PRs (30d)
- 1
Description
Hello! Thanks for amazing lib, its explanatory features are very helpful.
I'm trying to predict spikes of target metric during specific intraday events. It is clearly visible that spike occurs on the minute of the event, and the imact fades out during some time after it.
The problem that I'm facing is that the event may happen on different hours/minutes of the day, so custom daily or holidays seasonalities have problems with "learning" the time of the day when such spike is likely to happen.
Would you please recommend some workaround for that, so I can shift a time-window of such an event and/or to build seasonalities not attached to time of the day, but based on +/- timedeltas (hours/minutes) around an event?
Ideally, i would imagine sub-daily "holidays" functionality that does that kind of magic.
Contributor guide
Assessment
This issue has not been assessed yet.