Placeholder Map ticket: add animations to visualize changes over time

[danicahelb]

This is a new feature to be added to the stand-alone full screen map (and also map viz tool in the EDA). We would like to be able to "play a movie" to animate how some variable changes over time for points on the map.

This will be needed for our collaborations with WWARN, and should be delivered (on a dev site) by July 2023.

[danicahelb]

use cases:

• bubble markers over time would be useful for things like surveillance, like the new incident cases of malaria per 100K population, over time. for incidence, it would look at a snapshot of the bubble markers at each timepoint

• bubble markers over time would also work for cumulative cases of malaria, where instead of each timepoint being a separate snapshot, each timepoint would add to the magnitude of the bubbles in the previous snapshot

• bubble plots of proportions would allow users to select a single marker of resistance and see what percentage parasites have that resistance marker over time

• with WWARN we may want to see proportions of (the top 5? + other) resistance markers over time. consider a categorical color map with a lot more than 8 colors so that as new mutations develop and become predominant, they can be assigned a new color and replace one of the original mutations such that the markers only show 6 colors at any given time.

[danicahelb]

examples of animated maps:

bubble markers over time: https://vividmaps.com/indias-50-most-populous-cities-1901-2011/

donut markers over time: https://flightaware.com/miserymap/

this one is not automated, but we could use a different color bubble marker for each value and animate it over time: https://www.carbonbrief.org/mapped-how-the-us-generates-electricity/

https://blogs.sas.com/content/graphicallyspeaking/2020/10/16/animating-us-covid-19-data-over-time-on-a-county-map/

https://www.communityhealthmaps.org/blog/creating-a-covid-animation-through-time-with-qgis

https://vividmaps.com/percentage-of-white-people-in-us-fro/

[danicahelb]

Feedback from Ann & Danielle:

Some thoughts:

• The choice between point, line and choropleth map isn't really to do w timeseries. It's more about the nature of the data you have and what you mean to show or explore w it.
• Introducing new types of maps is likely costly so we should do it w careful consideration.
• It's nearly impossible to get a categorical colormaps w more than 8 colors unless we can broadly group the values were mapping the hues to. (Think grouping species by family or something, assigning hues to families and slightly varying the hue for each species.) Even that strategy has very real limitations though. And fwiw, people can't really make sense of more than about 7 colors at once anyhow..
• It's possible to map multiple related variables to a single point marker on a map. You could represent one by size, another color and a third opacity perhaps. You can also use a single summary value describing the relationship between two variables and use that instead (think correlation coefficient for ex). Again, there's a lot to consider if we go that route.
• Just bc we have geospatial data doesn't necessarily mean we need to use a map for everything. Maps are probably one of the most misused and overused viz tools on the planet.
• animations are super cool but if we can show the same information in a static plot that will be better for analysis. People are better at comparing two things next to each other in space (ex. two points on a line plot) than in time (ex. bubble size at time a vs size at time b)