lazd / coronaglobe Goto Github PK

Reading the ArcGIS blog it was noted that red might not the the best color for regions on the map (despite the ArcGIS map using red against a black map).

Other colors should be explored. This branch supports different colored heatmaps, but I couldn't find a gradient that looked right: https://github.com/lazd/coronaglobe/tree/heatmapcolor

Using population-weighted would give a more useful map, as discussed in the ArcGIS blog entry.

Population data could be pulled from GHSL.

It seems like the whole point of this visualization is to see how many people are sick - what areas are currently having to respond to the most active cases. Since "cases" is cumulative, it's more useful as a static representation once this whole thing is over. "Active" is more useful right now since recovered people are removed from the count.

An example difference right now is that in the March 23 data, South Korea has surpassed Wuhan in terms of active cases, but Wuhan still has 6x more "cases". South Korea should be getting more attention on the map.

Hi! I am trying to run coronaglobe locally and I am running into cannot find locations.json error. I don't see the file under coronadatascraper either. Am I missing anything?

Thanks,

A

As mentioned in the ArcGIS blog, it might be a better representation of the data if points were distributed randomly across the specified region instead of just one big point in the center.

It would look something like this, but with interpolation.

It should be possible to auto-step through the timeline with a button.

Right now, blob size and intensity is determined with a bit of a silly equation. It's logarithmic so things don't get too big or don't appear at all when too small, but it results in a disproportionately sized blob if there are too many points near each other (i.e. in Japan, where there are 3 points nearby):

It could look a lot better if the equation was adjusted.

It should be possible to search among the existing locations with an autocomplete.

Nice to have: Geocode the entered location if it doesn't appear in the autocomplete

Because Natural Earth's region for France includes French Guiana, randomly distributed points show up within it. This is incorrect.

https://cran.r-project.org/web/packages/viridis/vignettes/intro-to-viridis.html

Should use magma for bad things, green-yellow for recovered?

Right now, a new heatmap is generated each time the day switches. Instead, the cases should be interpolated between days.

Instead of clearing the heatmap on each frame between days, it would be ideal to add points to the map and update it, though I don't know how to handle removing points from the map in that case.

It could also be possible to fade between days.