[EPIC]: District Comparison

[NXXR]

Tooltip

Improve tooltip with a lens highlighting the selected disctrict and show scenario p25, p50, and p75 of selected scenario (and real value)

• Tooltip Example:
  • Lens highlights selected District (might take up too much space)
  • Left gradient shows scenario's p50 (center dot), p25 and p75 (gradient start/end)
  • Right shows tooltip text
  • Border colored depending on selected scenario
  • selected District colored in current real value (?)

Scenario Comparison

Allow scenarios to be compared in the district map

• display heatmap with relative difference between two scenarios or scenario and real value
• reduce districts to a node map with colored nodes
  • colored with a heat legend based on the differece between scenarios
  • colored with both scenarios as gradients from p25 to p75
  • Node with borders colored according to scenario and real value in center:
  • Node without borders:

[NXXR]

This paper: Comparing Performance Heatmaps by D. Krakov and D. G. Feitelson Discusses Comparison Techniques for Heatmaps (3 Dimensions of Data) and may provide a starting point for comparison techniques of the scenarios.

• Most straightforward solution would be to paint a heatmap of simple or ratio difference between both scenario values.
• Some relation to Kullback-Leibler Divergence may also be viable.
• The later comparison techniques (Ch 5) like quiver plot don't appear useful in our case

[Pawandeep-Kaur-Betz]

Heatmaps are the good solution for comparison. Very straight forward solutions would be, making a grid heatmap based on the geographical map. In this heatmap, rows are different simulations and columns are different states. Change in the color shows the change in the percentage of the infected people. Nested bars shows node information. For example like here. viz idea

[NXXR]

• move view to district map location
• 2D grid of nodes roughly georeferenced
• 2 views
  • scenario comparison
  • compartment comparison

[Pawandeep-Kaur-Betz]

For 2D grid of georeferenced nodes -

The solution needs to be implemented is available in paper https://www.gicentre.net/woodspatially2008

Inside the grids will be nested visualizations.

The code is old and is available in java with depedency to processing 3 https://www.gicentre.net/software#/treemappa/ https://github.com/gicentre/treemappa

It wont run in the current version of processing4. Needs to remove the dependancy on processing3 and clean the code. If the code works, then we can implement it in ts-js.

[Pawandeep-Kaur-Betz]

After the gridified map is finished, we might need the result of this study to use the effective zooming and lens style.

https://dl.acm.org/doi/abs/10.1145/3491102.3517673