I think it might be nice to show two examples here: one with samples for a univariate quantity (i.e., observational units are a single combination of location, origin date, horizon), and another with samples for a multivariate quantity (for example, we might get "sample trajectories" across horizons, where the observational unit is determined by a combination of location and origin date).
If we want to do something like this, the process of getting samples from the quantile forecasts in our existing example is demonstrated here.
I think it might be nice to show two examples here: one with samples for a univariate quantity (i.e., observational units are a single combination of location, origin date, horizon), and another with samples for a multivariate quantity (for example, we might get "sample trajectories" across horizons, where the observational unit is determined by a combination of location and origin date).
If we want to do something like this, the process of getting samples from the quantile forecasts in our existing example is demonstrated here.