This is motivated by the desire to more properly capture the NSSP dataset, which appears to report county-level information, but it is actually reporting HSA-level information (and repeating the same value for each county in an HSA).
To support HSA as a "geo_type" in the covidcast endpoint, the only absolutely required change would be to add "hsa" to the allowed geographies in the Covidcast CSV importer code:
|
# set of allowed resolutions (aka "geo_type") |
|
GEOGRAPHIC_RESOLUTIONS = {'county', 'hrr', 'msa', 'dma', 'state', 'hhs', 'nation'} |
It should be easy and helpful to also add sanity checks for HSA geo_values during acquisition.
Less easy but potentially even more helpful, we should also add server-side request marshaling/validation for HSA codes, but this needs GeoMapper support; see cmu-delphi/covidcast-indicators#2139 and note the discussion of "NCI Modified" HSA definitions, which are likely to be of consequence to this issue.
This is motivated by the desire to more properly capture the NSSP dataset, which appears to report county-level information, but it is actually reporting HSA-level information (and repeating the same value for each county in an HSA).
To support HSA as a "
geo_type" in the covidcast endpoint, the only absolutely required change would be to add "hsa" to the allowed geographies in the Covidcast CSV importer code:delphi-epidata/src/acquisition/covidcast/csv_importer.py
Lines 64 to 65 in 274cfe5
It should be easy and helpful to also add sanity checks for HSA
geo_valuesduring acquisition.Less easy but potentially even more helpful, we should also add server-side request marshaling/validation for HSA codes, but this needs GeoMapper support; see cmu-delphi/covidcast-indicators#2139 and note the discussion of "NCI Modified" HSA definitions, which are likely to be of consequence to this issue.