Hey, I'm experimenting with a timeseries that has hourly resolution with both daily and weekly seasonality, in addition to a slowly varying linear trend. I've been running into errors — LinAlgError: SVD did not converge — and am wondering if using a "long-seasonality" component for the weekly would be better and/or address the numerical issue. However, based on the docs, I'm having a hard time figuring out how to specify such a longSeason component.
I think including a full example in the docs would be hugely helpful! Or, if you could sketch out the expected code here, I might be able to put together a documentation example myself. Thanks in advance for your help!
Hey, I'm experimenting with a timeseries that has hourly resolution with both daily and weekly seasonality, in addition to a slowly varying linear trend. I've been running into errors —
LinAlgError: SVD did not converge— and am wondering if using a "long-seasonality" component for the weekly would be better and/or address the numerical issue. However, based on the docs, I'm having a hard time figuring out how to specify such alongSeasoncomponent.I think including a full example in the docs would be hugely helpful! Or, if you could sketch out the expected code here, I might be able to put together a documentation example myself. Thanks in advance for your help!