Computing different metrics is basically free and should help reveal different information. I will group these into four different formats - and each will have an issue. And a further issue to define the broad format that any metric function should follow. Any big picture things can be discussed here. The presumption is that any metric calculation will have access to a 'real' network and accompanying SWMM simulations for the same precipitation.
[x] #54
[x] #50
[x] #51
[x] #52
[x] #53
Note:
Based on my experience, KGE tends to give better results than NSE, have you tried it with KGE?
barneydobson
commented
6 months ago
Computing different metrics is basically free and should help reveal different information. I will group these into four different formats - and each will have an issue. And a further issue to define the broad format that any metric function should follow. Any big picture things can be discussed here. The presumption is that any metric calculation will have access to a 'real' network and accompanying SWMM simulations for the same precipitation.
Note:
_Originally posted by @cheginit in https://github.com/ImperialCollegeLondon/SWMManywhere/pull/60#discussion_r1514805016_