Skip to content

Conversation

@jonasbhend
Copy link
Contributor

@jonasbhend jonasbhend commented Oct 21, 2025

To avoid comparing apples with oranges, we need to make sure that the exact initialization specified in the config are used for evaluation for both the forecasts and baselines. This was previously not the case for the baselines which used the nearest available initialization in the dataset.

Changes

  • exact selection of ref_times in baseline
  • fix configs for availability of COSMO-1E baseline (starting in 2020-08-01)

@jonasbhend jonasbhend requested review from dnerini and frazane October 21, 2025 09:57
@jonasbhend
Copy link
Contributor Author

Checked all example configs, good to go.

@jonasbhend jonasbhend marked this pull request as ready for review October 21, 2025 11:25
@dnerini dnerini merged commit 67b1906 into main Oct 21, 2025
4 checks passed
@dnerini dnerini deleted the MRB-550-Inconcsistent-forecast-initializations-in-evalml branch October 22, 2025 04:52
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants