hubverse-org / flusight_hub_archive

Hubversion of FluSight 1 (2015-2019)
MIT License
2 stars 1 forks source link

No model metadata #13

Closed lmullany closed 1 week ago

lmullany commented 2 months ago

I believe there are no model metadata files (perhaps I need to do a more thorough search of the original repo and documentation). As a result the model-metadata folder is empty. This means that any external functionality that depends on information in this folder would not work, or would need to be adapted to account for situations where model-metadata is missing

lmullany commented 2 months ago

Among numerous implications is that we directly know if a team/model is an ensemble of other models, or is a secondary rather than primary model, etc.

nickreich commented 2 months ago

In correspondence with CDC, it looks like there might exist some doc/pdf files that have narrative descriptions of the model. Seems like it might be good to try to capture some basic metadata fields for each model, even if it means some painstaking manual entry. Here's a start at a possible list, following the recommended standards, where * indicates required:

nickreich commented 2 months ago

Model Metadata now available here: https://github.com/cdcepi/FluSight-forecasts/tree/d6176ddf0426ba66662f2fd5e056a8fde76362d8/Team%20Methodologies

lmullany commented 2 months ago

okay, that's helpful - at first glance, this does look painstaking, but could be an important process to go through.. Not sure that it looks very complete though. I'll take a closer look.

lmullany commented 1 month ago

Melissa Kerr and Martha Horn have been working on creating yaml files for each of the team-models. A push of these yaml files to the model-metadata folder will be part of the solution to this issue

lmullany commented 1 week ago

Closed via #33