<-- Advertise here.

Cartographer Zachary Forest Johnson has presented a novel way of representing a large number of geo-located points on a map. He took inspiration from the original 'hexbin' visualization method, short for hexagonal binning (PDF!), which originally is meant to aggregate scatterplotted points into an hexagonal grid, of which each cell is then colored or sized according to a specific value (e.g. relative density or average numerical value) of those points that fall within it. This method avoids the need to render thousands of, potentially overlapping, points (e.g. see NYTimes Maps U.S. Census Data), hereby reducing visual clutter and rendering complexity, in particular for real-time applications.

The new possibilities of the HexBin are demonstrated through the representation of all Wall Mart locations in the US [github.com].

HexBin is now available as a D3.js function, and can be combined with PolyMaps to make hexagonal renditions of geolocated points. Some history about HexBin and much more detailed information about its geographical implementation can be found here.


It's nice, but the obvious limitation is that areas which are high density with but just happen to be split over two hexagons would appear lower than otherwise. Similarly areas with high densities that fall completely in one hegacon would appear higher than others even though it may have a similar density.

Should work ok with data that don't have such high peaks though.

Fri 21 Oct 2011 at 11:34 AM
Commenting has been temporarily disabled.