Cuong Duong

If you're on Windows, you can either: * Download the latest binary wheels here: https://github.com/facebook/prophet/actions/runs/2275140778 (we're working on releasing this to PyPI but they need to be downloaded manually for...

This is awesome! I'm keen to try it out. It seems that the confidence intervals using NumPyro are wider compared to using Stan - do you know why this is?...

Just an update: the CI checks all passed (installing the numpyro dependency and running the numpyro test script), so I'm pretty confident that this works properly. I'll try to do...

Hey @freddyaboulton @nicolaerosia ! Finally getting back to this one. We're at a place now where `stan` installation isn't as much of an issue, but I think the benefits that...

Could I ask what command you're using to install prophet? This should work with pip: https://github.com/facebook/prophet#installation-in-python If you're looking to build the dev version it's best to follow these instructions:...

Hmm unfortunately I don't think we host the documentation for the historical version.

Hmm if you're using the default `rstan` backend, the model should already be compiled as a part of installing the package. I'm not sure if there's an issue with calling...

See https://github.com/facebook/prophet/issues/2107#issuecomment-1133927696

You could try to increase the Fourier order for time-of-day ("daily") seasonality (more Fourier terms = more flexible fitting): https://facebook.github.io/prophet/docs/seasonality,_holiday_effects,_and_regressors.html#fourier-order-for-seasonalities

Sorry all for the delay on this, had the chance to look at it more deeply this weekend though and after making some tweaks, I think it's ready to merge!...