Mapping Coronavirus

The three interactive plots shown below were all generated using the associated github repo https://github.com/EricThomson/covid-mapping. At the end I describe the workflow for developers who want to work with this dataset but are not sure where to start.

The data, updated nightly, are from the generous folks at https://covidtracking.com/.

Cases in the US versus Date

The dropdown menu in the graph lets you switch between logarithmic and linear scale for the y-axis.

Heat-map

The colors are drawn after a logarithmic transform.

Bubble map

This one is pretty hard to scale correctly: it might be good to switch to a logarithmic scale so that New York isn’t so huge compared to the others. Let’s put that on the to-do list.

Workflow

For Python peeps, you can find the Jupyter notebook I used to generate the above plots here:

https://github.com/EricThomson/covid-mapping/blob/master/covid_map.ipynb

The notebook includes how to download the data, which is conveniently available online as csv files and updated nightly. Within the notebook, we render the interactive graphs using plotly (https://plot.ly/), save them as html files, and then embed them in this web page using frames.

This web site is hosted by GitHub. Basically, make a docs folder in your repo, throw your html files and index.md in there (and pick this option in settings, as well as a theme), and github will tell you the url to point people. Styles changes by adding stuff you will find in the assets folder. The web site was set up by repo contributor Julia https://github.com/juliakm. Thanks Julia!

One slightly annoying wrinkle: the Jupyter notebook needs to be run each night to pull the fresh data and update the plots. Currently I am doing it by hand: this could all be ported to a server using papermill and a scheduler (e.g., cron). For production-ready code we likely would just use a py file not a notebook.