We would like to be able to compute point forecasts for quantile- and sample-based forecasts. The cleanest way to do this would be to have a function that converts the forecast into a point forecasts and then you can score it again.
Something like:
Additional considerations:
For sample-based forecasts that would be quite similar to summarise_scores(data, fun = median). For quantile scores passing in your own function maybe makes less sense as there is not much you can do except the median. (see discussion in #340.
Meta-issue: #433. Next issue: #434
We would like to be able to compute point forecasts for quantile- and sample-based forecasts. The cleanest way to do this would be to have a function that converts the forecast into a point forecasts and then you can score it again. Something like:
Additional considerations: For sample-based forecasts that would be quite similar to
summarise_scores(data, fun = median)
. For quantile scores passing in your own function maybe makes less sense as there is not much you can do except the median. (see discussion in #340.