hubverse-org / flusight_hub_archive

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

team and model names across season #12

Closed lmullany closed 1 month ago

lmullany commented 2 months ago

Some teams changed their name from year to year, and thus a single team-model can have multiple folders in the model-output folder. If this change in team-model differs only trivially (for example, spelling name change, or writing our the model name instead of acronym), rather than a substantive change in the model itself, then perhaps we should work towards consolidating these multiple folders

lmullany commented 1 month ago

Question Raised (@bsweger): Should the model output folder names allow spaces? If yes, then if that folder name is incorporated into the name of the files contained within that folder, those filenames will also contain spaces. If that is not okay, then we need to either not allow spaces in the folder name, or replace the spaces in the filename with underscores

bsweger commented 1 month ago

My .02 (and I raised an issue over in the hubverse-transform repo): any cloud-based data transforms should handle all incoming file/folder names gracefully.

To date, that transform module has assumed that incoming data has gone through the hubverse model-output validation function, but if we want to use it with converted archived data, it should handle these types of edge cases.