Cuong Duong
Cuong Duong
The code in the docs will work if you keep the same number of parameters. By adding 4 regressors, you're adding 4 extra parameters, so if you'll need to pass...
See https://github.com/facebook/prophet/issues/2107#issuecomment-1133927696
Yeah I think you either want to model each store separately, or have multiple regressors, one for each store (although I'm guessing this will lead to overfitting).
holidays>=0.14.2 should have fixed this issue.
I don't mind this feature, although I'd like it to raise a warning saying we have filled the extra regressor columns with, e.g. their average historical value (or, we can...
Sure thing :) Feel free to raise a PR and I'll take a look
Prophet does [detect trend changes](https://facebook.github.io/prophet/docs/seasonality,_holiday_effects,_and_regressors.html#fourier-order-for-seasonalities) automatically. If there are historical periods you know won't be repeated in your forecast horizon, you can also flag them as [one-off holidays](https://facebook.github.io/prophet/docs/seasonality,_holiday_effects,_and_regressors.html#modeling-holidays-and-special-events) to help...
Hmm I'm not sure what the issue is here, it seems like it could be platform-specific and only for an older version (0.7.1). Can you try installing `pystan 2.19.1.1` and...
@andrjohns not sure if you ever saw this warning: ``` 23:23:47 - cmdstanpy - WARNING - Non-fatal error during sampling: Exception: normal_id_glm_lpdf: Scale vector is 0, but must be positive...
When you do `.astype('