Allow targets to have an ensemble dimension in UnbiasedEnsembleMeanSquaredError.
#244
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…quaredError`. If present, debias the target contribution as well. One might ask, "how do you get an ensemble dim in the target?" Very good question, here is an answer. You could use the child CLs to... 1. `StackToRange` the 'ensemble' and 'lead_time' dims in predictions into a new ensemble dim 2. `Rename` the target's 'valid_time' dim into an ensemble dim 3. Use this function This same "trick" could be used for all ensemble metrics...effectively turning them into distributional distance measures. E.g. we could do a version of CRPS that subtracts (1/2) E|Y - Y'|. PiperOrigin-RevId: 738903556
copybara-service
bot
force-pushed
the
test_738903556
branch
from
e0dbc33 to
c003573
Compare
copybara-service
bot
changed the title
UnbiasedEnsembleMeanSquaredError.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Allow targets to have an ensemble dimension in
UnbiasedEnsembleMeanSquaredError.If present, debias the target contribution as well.
One might ask, "how do you get an ensemble dim in the target?" Very good question, here is an answer. You could use the child CLs to...
StackToRangethe 'ensemble' and 'lead_time' dims in predictions into a new ensemble dimRenamethe target's 'valid_time' dim into an ensemble dimThis same "trick" could be used for all ensemble metrics...effectively turning them into distributional distance measures. E.g. we could do a version of CRPS that subtracts (1/2) E|Y - Y'|.