Bad data for Solano county cases on 2020-09-02

[Mr0grog]

The timeseries Solano county cases has an erroneous, huge bump in the data on September 2, 2020:

{
  "cases": [
    // ...
    {
      "date": "2020-09-01",
      "cumul_cases": 5547
    },
    {
      "date": "2020-09-02",
      "cumul_cases": 25806
    },
    {
      "date": "2020-09-10",
      "cumul_cases": 5778
    },
    // ...
  ]
}

This is coming from the underlying data source (https://www.arcgis.com/home/webmap/viewer.html?url=https://services2.arcgis.com/SCn6czzcqKAFwdGU/ArcGIS/rest/services/COVID19Surveypt1v3_view/FeatureServer&source=sd):

We should figure out a way to smooth this out, whether by hardcoding a manual correction for that date, pulling data from the state, or something else.

[Mr0grog]

Looks like we have this problem for deaths and tests, too:

{
  "tests": [
    // ...
    {
      "date": "2020-09-01",
      "cumul_tests": 75559
    },
    {
      "date": "2020-09-02",
      "cumul_tests": 188850
    },
    {
      "date": "2020-09-10",
      "cumul_tests": 79112
    },
    // ...
  ],
  "deaths": [
    // ...
    {
      "date": "2020-09-01",
      "cumul_deaths": 47
    },
    {
      "date": "2020-09-02",
      "cumul_deaths": 106
    },
    {
      "date": "2020-09-10",
      "cumul_deaths": 48
    },
    // ...
  ]
}

[Mr0grog]

And this isn't showing up in their dashboard, which is based off this data:

It looks like it’s reading a different field than we are:

• Chart on dashboard: Number of specimens collected on this date
• Us: Cumulative number of cases on the date reported to Solano public health

So we should maybe just be looking at different fields. That doesn’t solve the issues for deaths and tests, though. ¯\_(ツ)_/¯