Friday, September 7, 2012

Sparkline Maps

A while back I was working with a colleague, Leo Dion, on a side project around visualizing buckets and buckets of Bureau of Labor and Statistics data.  Specifically, we wanted to show reams of historic price information for various commodities at various locations.
The result was a Tufte sparkline pin that danced with relative proportion as the data play through.  It looked a little like a flashmob of oscilloscopes doing the worm.  Here are a few screen-captures from an application made for Microsoft's Digital Crimes Unit that used this method to show regional trending of botnet attacks animating through time.

The point of a sparkline is to show potentially large amounts of data without getting bogged down in the nitty gritty of axes and labeling.  The trade-off for the reduction in specifics is a data-dense visual that offers an immediate sense of variability in historic activity. This same amount of data in spreadsheet form is a much different animal (with it's own benefits and pitfalls.  Namely, more specifics but slower 'aha's).  The first two snapshots happen to show 24 hours of data at the country level.  When I reduce the time slider to something like 2 or 3 hours the charts reduce to a very basic change indicator of either stasis, increase, or decrease.  When seen animated throughout the archive, the effect is pretty arresting as the tracks race up and down.

Why a map?  These guys could perform just as well in an alphabetical list of countries, but when viewed in this context additional dimensions can come into play like a referential overlay of national corruption indices (not shown), digital infrastructure, or just their relative proximity one to another.

And all that stuff.

No comments:

Post a Comment