Texas county data

[sglyon]

Some discrepancies between our Texas county data and TX county data from other sources seems to exist.

Our data comes from the state dashboard here: https://txdshs.maps.arcgis.com/apps/opsdashboard/index.html#/ed483ecd702b4298ab01e8b9cafc8b83

Looking at Ector county (for example):

cmdc-tools on  master [$!] via 🐍 v3.8.5 via C base on ☁️  us-east-1 
❯ http "https://api.covidcountydata.org/covid_us?location=eq.48135&variable=eq.cases_total&order=dt.desc&limit=1" | jq
[
  {
    "dt": "2020-08-17",
    "location": 48135,
    "variable": "cases_total",
    "value": 2547
  }
]

And the dashboard shows:

Looks like we match. However, if we look at the NYT data we get something different:

cmdc-tools on  master [$!] via 🐍 v3.8.5 via C base on ☁️  us-east-1 
❯ http -d --body "https://raw.githubusercontent.com/nytimes/covid-19-data/master/live/us-counties.csv" | rg 'confirmed|Ector'
Downloading 153.17 kB to "<stdout>"
 | date,county,state,fips,cases,deaths,confirmed_cases,confirmed_deaths,probable_cases,probable_deaths
2020-08-17,Ector,Texas,48135,3771,59,2547,59,1224,
Done. 153.17 kB in 0.02435s (6.14 MB/s)

Looks like the 2547 variable is NYT's confirmed_cases, which matches our cases_total and the state dashboard.

However, NYT also has the very different 3771 value for cases.

Thoughts on what we should do here?

cc @TomGoBravo

ref act-now-coalition/covid-data-model#616