BUG: change of internal and external climatology
Bug
Error description
When calculating the climatological skill score, the data sets are interchanged for accident. Clearification:
internal means that the data is already known to the model -> train and val
external means that the data is not known to the model and a good skill score would be harder to archive (but also will reward a forecast more= -> test
Error message
silent error
First guess on error origin
Error origin
Solution
-
use train_valset as internal climatology -
use testset as external climatology -
update required in any description? -
confirm also that skill scores are calculated on testdata
Edited by Ghost User