I like the idea of the areas being somehow proportional to the variable. When two (or more) circles intersect we loose some visual effect ... Neither darker coulour on the intersection nor numbers representing the actual value of the (sum of) the variable offer, in my opinion, a neat solution. Kriging may help in some cases just as redistributing points at random within a finite surface: but neither of the two are totally satisfactory, as they tend to assume continuity of data that is not necessarily in there to be seen; both fail to render a true visual image of concentrations occurring in small areas. I had been thinking about circle coalescence: drops of a viscous liquid getting that bean-shape as they touch each other and progressively merge as you add more drops. If only I could find the additional quantity r to buffer each circle of radius R so that the additional area of the enlarged circle (now of a radius= R+r) would compensate for the area lost by that circle due to the intersection... At panned zooms, all would merge into a single circle, which would then acquire bean-shaped forms before individual circles/points would start appearing as you zoom in. But my skills don' allo to do that myself Unfortunately in geometry and in Manifold Scripting do not enable me to grasp how to achieve that... FWIW here is the source of COVID-19 data I use for my work: https://opendata.ecdc.europa.eu/covid19/casedistribution/csv
s |