US Counties — EndCoronavirus.org

US COUNTY-LEVEL MAPS and classification of COVID-19 cases

Created by: Olha Buchel from the New England Complex Systems Institute and Joseph D. Ortiz from Kent State University, Department of Geology.

Code to generate the interactive maps at https://www.endcoronavirus.org/us-counties

Web site setup and data flow

The page https://www.endcoronavirus.org/us-counties is hosted by SquareSpace. It embeds https://obuchel.github.io/classification/classification_map.html in an IFrame.

In the source code of that HTML page you find several data sources

fetch('classification/classification_ids_counties2.json')

fetch('counties.json')
 
 if (["630","16","316","580","850"].indexOf(num)>-1) {
    
    var url='classification/data_counties_'+num+'.json';
} else {
    
    var url='classification/data_counties_840'+num+'.json';
}
fetch(url)

classification/classification_ids_counties2.json

Has records like

{
    "n": "Cuming, Nebraska, US",
    "id": 84031039,
    "v": 0.8484848484848485,
    "c": "red",
    "max": 22
}

counties.json

This is a shapefile (JSON format), with records like

{
  "type": "Feature",
  "geometry": {
    "type": "Polygon",
    "coordinates": [
       "lots of lon-lat pairs."
    ]
  },
  "properties": {
    "STATEFP": "31",
    "COUNTYFP": "039",
    "COUNTYNS": "00835841",
    "GEOID": "31039",
    "NAME": "Cuming",
    "NAMELSAD": "Cuming County",
    "LSAD": "06",
    "CLASSFP": "H1",
    "MTFCC": "G4020",
    "CSAFP": "",
    "CBSAFP": "",
    "METDIVFP": "",
    "FUNCSTAT": "A",
    "ALAND": 1477895811,
    "AWATER": 10447360,
    "INTPTLAT": "+41.9158651",
    "INTPTLON": "-096.7885168",
    "value": 2
  }
}

From here we can figure out some relevant fields:

"GEOID": "31039", this matches the "id": 84031039 shown in classification_ids_counties2.json above.
STATEFP is the State FIP code
value: not sure yet where that comes from, and what it is used for

'classification/data_counties_'+num+'.json'

Data for a given county, indexed by its GEOID, prepended with '840' (except for a handful of counties.) These are time series for the county, here are the most recent 4 values for Cuming County:

{
  "dates": ["5/19/20", "5/20/20", "5/21/20", "5/22/20"],
  "max_14": 22,
  "max": 32,
  "value": [1.55, 1.5, 1.45, 1.4 ],
  "time": [ "5/24/20", "5/25/20", "5/26/20", "5/27/20" ],
  "original_values": [0, 0, 0, 0]
}

My interpretation for the fields:

dates is the x axis for the time series
max_14 maximum case count last 14 days
max maximum case count
value smoothed case count
time looks redundant
original_values the case counts

Data flow on the python side

The script prepare_classification_counties_final.py is used to write the JSON files for the website. Just like with the JavaScript on the web page, you can look for open() in the python file.

my_ids.json

This looks unused.

Reading CSSEGISandData

https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_US.csv')

Writing: 'data_counties_'+str(ids[recs.index(name)]["UID"])+'.json'

These are the files with the county data.

Writing: data_counties.json

This is a list of records like this:

{
    "color":color,
    "province":name.split(",")[0],
    "country":name.split(",")[1],
    "id":"new_id_"+str(ind4),
    "value1":ratio, 
    "dates":tim2,
    "value":y3
}

presumably one entry per county. This file is apparently included in the github repository as classification/data_counties.json.

Writing: classification_ids_counties2.json