Background
The first iteration of the hubverse-transform function copied logic from the Hubverse validation function to get round_id, model name, and team name.
However, the ongoing work to convert archived hub data to Hubverse format resulted in some good conversations about these data, so hubverse-transform needs to be updated to reflect the resulting decisions.
Required work
Update the hubverse_transforms module to add 2 columns to incoming model-output files instead of 3 (model_id and round_id instead of model_name, team_name, and round_id).
Definition of done
[x] hubverse_transforms adds two columns to incoming model-output files submitted via cloud-based hubs: model_id and round_id
Note: Until we make a decision otherwise, this code will continue to assume that hub round_ids are in YYYY-MM-DD format). Our first big test of the transforms will be on the incoming FluSight archives, and those model-output files use YYYY-MM-DD, so we're good for now.
[2024-05-03: updated ticket based on recent convos/decisions that emerged from @lmullany's work to convert the archived FluSight hub to Hubverse format: model-id = model name + team name and is round-id required to be in yyyy-mm-dd format.]
Background The first iteration of the
hubverse-transformfunction copied logic from the Hubverse validation function to get round_id, model name, and team name.However, the ongoing work to convert archived hub data to Hubverse format resulted in some good conversations about these data, so
hubverse-transformneeds to be updated to reflect the resulting decisions.Required work Update the
hubverse_transformsmodule to add 2 columns to incoming model-output files instead of 3 (model_id and round_id instead of model_name, team_name, and round_id).Definition of done
hubverse_transformsadds two columns to incoming model-output files submitted via cloud-based hubs:model_idandround_idNote: Until we make a decision otherwise, this code will continue to assume that hub round_ids are in YYYY-MM-DD format). Our first big test of the transforms will be on the incoming FluSight archives, and those model-output files use YYYY-MM-DD, so we're good for now.