kge
kge copied to clipboard
Added a checking to see whether any values are NaN in the scoring.
Regarding issue #220 - by adding a check if there are any NaN values, a ValueError will be thrown if a nan-value is found.
If I am to be self-critical, there is a risk that the speed might be lower, but I still think it is necessary in order to ensure a correct evaluation.
I agree that it's good to have a check for NaN values. It should be in the evaluation code, not the model code, however.
Good point. I will make changes accordingly!