search

facebook / prophet

Tool for producing high quality forecasts for time series data that has multiple seasonality with linear or non-linear growth.
https://facebook.github.io/prophet
MIT License
18.08k stars 4.49k forks source link

Is FBProphet appropriate for modeling percentages or probabilities? #2463

Open lizre opened 1 year ago

lizre commented 1 year ago

Hi! Is FBProphet appropriate for modeling percentages or probabilities? What about skewed data, like counts? ❤️

AlexandroLuis commented 11 months ago

The model is a good fit for Probabilities. with the right parameters it can be very good in it, try adjusting seasonality_prior_scale=0.7, # Adjusted value changepoint_prior_scale=8, # Adjusted value These two are used to control the flexibility of the model and how much it emphasizes capturing seasonality and changepoints in the data. And the seasonality to make the series detect patterns of seasonality model.add_seasonality(name='monthly', period=30.5, fourier_order=6) # Adjusted Fourier order