Group data points around the largest points.
The recipe is as follows:
1) sort data points according to their sizes (here size = d[3])
2) use top 10% points as Voronoi sites
3) bin all data points according to their Voronoi cell, using voronoi.find(). The binning is rendered by using the Voronoi sites’s color.
See also the static version — much less code to parse!
Original work by Philippe Rivière for d3-voronoi (issue 17).