Update: more specifically, as discussed in #515: Remove any internal handling of an attribute and simply throw a warning.
validate(example_quantile) |> attributes() produces an item
$warnings
[1] "Some forecasts have different numbers of rows (e.g. quantiles or samples). scoringutils found: 1, 23. This may be a problem (it can potentially distort scores, making it more difficult to compare them), so make sure this is intended."
The warning comes from the fact that NA values in the example data aren't filtered out. We should
- make sure that
NA values are filtered out to get rid of that warning
- make sure that if an actual problem is found, a proper warning should be thrown. Same for messages.
Update: more specifically, as discussed in #515: Remove any internal handling of an attribute and simply throw a warning.
validate(example_quantile) |> attributes()produces an itemThe warning comes from the fact that
NAvalues in the example data aren't filtered out. We shouldNAvalues are filtered out to get rid of that warning