A highly-performant, mobile-friendly interactive timeseries visualizing the spread of COVID-19 in the United States.
- Novel. Visualize derived county-level active cases (per capita) on any day between
Jan. 22, 2020andMar. 9, 2023. - Fast. Thanks to instancing techniques and a hardware-accelerated
ctx.drawImage(), rasterization of the map and datapoints only happens once – giving you a fluid, jank-free experience even when zooming or panning. - Compare. Click on any county to open a chart visualizing its data over time. Select multiple counties simultaneously and compare them within a shared, synchronized date range.
- Mobile-friendly. Full touch gesture interactions for both map & charts.
- Share. Copy a URL to the clipboard linking directly to the visualization on a selected day.
The John Hopkins University COVID-19 Dashboard stood as a source of truth for many of us following the COVID-19 pandemic, right up until the university's cessation of data collection on March 10, 2023. Their dataset, archived on GitHub, contains total cumulative cases per county in the United States. When visualizing this data over time, it quickly resembles the shape of a population map – providing little value or insight with regards to the movement of the virus.
This visualization attempts to augment the JHU dataset in two ways:
- The inclusion of county-level census data for visualizing data per-capita.
- Visualizing active cases instead of cumulative cases. In the absence of a reliable dataset for active cases on a county level, the values are calculated based on the assumption that it takes, on average, 21 days to recover from COVID-19 (e.g.
Σ(i = n-21 to n-1) Δi).
These two considerations enable us to visualize the shape of COVID-19's movement, even if the values themselves are derived. The following are some interesting examples of some significant regional hotspots: