Standardized data models

[jjackson12]

Right now, the person and/or household models are not uniformly defined. It's painful having to map field names from module to module, and now I need to add a field to one place (vr_program_id) and that will require modifying several components in unintuitive places.

There should be a standardized Person model and a Household model. These should be passed around instead of miscellaneous objects/json. This will make life better. There will be a models.py file that is shared by all modules. It can translate the schema from different formats (BigQuery, NCOA data, Mailer formatting, etc) all in one place, and adding/removing output data will be easy since it'll all be stored in the same object.

There also should be a List model that is built around a collection of Persons, that can generate output lists as well as handle functionality for filtering and more.

[jjackson12]

This could be a good way to help facilitate this other action item: "For all fields, make the multi-select fields reference the unique values in the dataset so far (though this is computationally expensive, but not so much if I can do it from a bigquery call rather than referencing the whole dataframe)" --> The List model–with a cached object in the frontend–can handle this in an abstracted way.

[jjackson12]

Update: Working on this for a little bit in a separate branch, I'm not sure the juice is worth the squeeze. Can come back to it later if it feels it is worth it