Investigate fallback for sparse tracts

[emma-nechamkin]

for tracts that are missing a lot of data (like the pine ridge tracts), investigate impact of using a single criterion to assign disadvantage.

[emma-nechamkin]

add a category for data availability itself, or lack thereof

[emma-nechamkin]

add a category for data availability itself, or lack thereof

[emma-nechamkin]

add a category for data availability itself, or lack thereof

[emma-nechamkin]

Working on this now

[emma-nechamkin]

There actually aren't that many of these tracts! A lot of these tracts are mostly water -- if we restrict to just land area > water, we get very few tracts with >=2/3 of indicators as NaN (I get 60 outside of PR, MP, GU, AS, USVI)

[emma-nechamkin]

I think there might be a bug? We should see the forbidden island here: 15007041200. We need to be doing this, I believe, after the geo merge -- but we don't have an easy way to do so. There might be a difference between "not in the data" and "not even included in the stuff that could be the data"