Data Sources
The Coronavirus Mapping Project is sourced using a combination of public data from state health departments and public maps from ArcGIS. The data has been aggregated, indicators calculated, and joined with census data and stored in our data platform Standard Data here. This data is free for anyone to use for whatever purpose they deem necessary. It is available as a CSV download as well as through a JSON API.
Map building and Data Visualization with Metabase
If you've been following @tjmule's twitter stream, you know the Standard Co / Secure Data Kit team has been leveraging Metabase and maps in general to try and tell a more visual story of the #coronavirus outbreak. There are plenty of maps out there (in particular this one from Johns Hopkins) but what most of them lack is a little more specificity. We're in particular interested in county-by-county maps that share the status of the outbreak and how that data might drive awareness and inform our behavior locally.
It's with that in mind that we're creating this exploration to start to compile resources we use to visualize and track the spread of COVID 19. Specifically state by state county geoJSON sources, data sources of the outbreak, etc. This is a big effort so we're starting small and local but will hope to grow this in the coming days and weeks.
Data Management Tool
We use Secure Data Kit (duh). It allows us to easily spin up new data collection forms (for web and native mobile). If you want to set up a form to collect your own data, go for it -- we have a freemium version.
Mapping Resources
We're not native map makers -- we're web developers. So geoJSON is our preference for building maps. And it's super quick and easy.
First we wanted to build a visualization (updated 12 March 2020) that focused entirely on Africa. To do that, we needed regional boundaries. I used The WHO / AFRO boundaries found here (I ended up tweaking it a bit).
We've since created geoJSON of each US State down to the admin 2 (county) level. We used the FIPs code system for identifying states / counties so we would have a foreign key to other systems. I only mention this because up until recently, I had no idea what the FIPs system was. When we have a spare moment (in short supply right now!) we'll add links to each geoJSON. We found those particularly tricky to find so we want to make that resource freely available.
Data fragmentation and County Level Data
On March 13th - March 16th, we took up the manual and experimental effort of looking at county websites for county level data. Could we round up the data by hand? (yep), Could we load it into SDK? (sure, we do that every day), and could we build a dashboard that people would find useful? (we thought so). Everything was very informal and housed on www.standardco.de
A Richer Data Set
On March 17th, We rolled out a website that consolidated the dashboards we'd created into a single place. Our data efforts were still largely manual. We relied on the the talented folks at The COVID Tracking Project for state level case numbers (published everyday at 4pm PT).
We relied on our own data entry for county level data. This was...tedious to say the least. We compiled a list of states and listen to when they update their data (visible as a google doc here). These links now appear at the state level view.
We had written a few simple scripts to compile some basic stats like percent change day to day. And we continued to upgrade the dashboards to be more informative.
What's next? Get Involved!
All of a sudden, a few really well done data resources have come online. We're looking to aggregate these resources, add color, and make this data more accessible at the local level. If you’re interested in contributing or learning more, please reach out to TJ Muehleman on the Standard Co team (tj@standardco.de).