statsforecast
statsforecast copied to clipboard
[FEAT] Add `quantile` argument to the `StatsForecast.forecast` method
Is your feature request related to a problem? Please describe. It's not entirely straightforward to transform prediction intervals into quantiles.
Describe the solution you'd like
The argument quantile. It should receive a list of floats between 0 and 1. The levels associated with those quantiles should be computed inside the forecast method. Then the final data frame should contain specific columns for each model and quantile (model-q-{quantile}). Therefore, a user could compute prediction intervals and also quantiles at the same time. Also, it will be possible to calculate only quantiles.