Dorling Cartogram of Unemployment, 1980 – 2009

If you wanted to create web-based interactive visualizations prior to a year or so ago, you were pretty much limited to manually creating it  in Flash. Times are changing. Fast. Even outside of Tableau Public, ManyEyes, and the other great (and often free) web-enabled viz tools, you can now deliver really advanced graphics using javascript and a scalable vector graphics (SVG) enabled browser. A recent showcase of what you can do opens a window into where web-based visualizations can (and will) go.

Just to show how easy it can be to create a relatively complicated visualization, I’ve taken one of the examples presented in the showcase, and changed the data set. As the code is still being developed, I had to hack some of the code to work with the new data, but in the future I’m sure it will become much easier. Anyway, onto the example.

Dorling cartograms vary the size of a geographical region depending on a measure. They work well when you want to show a time history of regional variations in the measure. Why are they better than a just series of sparklines or a choropleth? Firstly, there may be trends larger than a state that add understanding to the data – for example, does a trend start in the southeast, and spread to other parts of the county? This would be hard to see on state-based line graphs.

Choropleths, where each geographical region is shaded to encode the measure can work well when animated to show the spread of a trend, but small geographical areas (I’m looking at you Rhode Island) can be almost impossible to read. This is where the Dorling cartogram comes into it’s own with some datasets. It can still suffer from smaller geographical regions being hard to see, or large ones dominating the view, depending on the variation in your measure. They also suffer in that same way that pie charts do – humans find it difficult to accurately compare 2D areas – but they do work very well on giving you a sense of how the data flow and regions change over time.

I’ve used the unemployment rates from 1980 to 2009 to produce this Dorling cartogram. Let the map settle for a bit – it may take up to 20s on slow computers/browsers. and then click the play button. I think it shows the ebb and flow of unemployment well over the three decades very well.  If it doesn’t show, chances are, it’s your browser (Firefox sometimes has problems, Chrome seems fine, IE8 and before won’t work). If you have a non-compliant browser , I’ve embedded a video of the animation.

Do you think your business would benefit from seeing information in new ways? If you’re stuck with row upon row of data in Excel, let Data Driven provide new insight and help you manage your business better.

13 Comments


  1. Alex Kerin
    May 22, 2010

    Chrome and Safari work well, Firefox (at least my version) is buggy – working with the developers to understand why.


  2. epicsystems
    May 25, 2010

    Dear Sir,

    I have the pleasure to brief on our Data Visualization software
    “Trend Compass”.

    TC is a new concept in viewing statistics and trends in an animated
    way by displaying in one chart 5 axis (X, Y, Time, Bubble size &
    Bubble color) instead of just the traditional X and Y axis. It could
    be used in analysis, research, presentation etc. In the banking
    sector, we have Deutsche Bank New York as our client.

    Link on Chile’s Earthquake (27/02/2010):

    http://www.epicsyst.com/test/v2/EarthQuakeinChile/

    This a link on weather data :

    http://www.epicsyst.com/test/v2/aims/

    Edit: Lots of links follow..


    • Alex Kerin
      Jun 02, 2010

      Thanks for the info – I trimmed your reply a little to scrub out some of the links. The software looks great, works smoothly. One downfall is that it’s flash, so as the use of flash changes (read Apple devices), SVG and javascript may become the way we do this. However, this method isn’t near ready for primetime yet, so flash it is.

  3. [...] tools continue the possibilities for home spun data visualizations using javascript. See my Dorling Cartogram using Protovis for some really advanced stuff. Category: Data Visualization, [...]

  4. [...] Kiren's description of his work here.   Posted 06-04-2010 5:32 AM by Graham Griffith Filed under: jobs, unemployment, graphic [...]

  5. [...] discussed recently about how new javascript libraries (especially Protovis) have the potential [...]

  6. [...] DataDriven, comes this interesting animation of the historical state unemployment [...]

  7. [...] DataDriven, comes this interesting animation of the historical state unemployment [...]

  8. [...] DataDriven, comes this interesting animation of the historical state unemployment [...]

  9. [...] Data Driven Consulting provides a dynamic visualization of national unemployment by state from 1980 through 2009. [...]

  10. [...] Unemployment Cartogram 1980-2009 ( Hat tip: The Big Picture) [...]

  11. [...] getting them aligned and not messing up line heights is generally annoying. As discussed in the Dorling Cartogram, Protovis Dashboard, and Google Intensity Map posts, the options for generating graphics are [...]

  12. [...] when trends are shown. Imagine trying to understand employment without information on what happened last year, last administration, or last recession. All of the reports that have predecessors should show [...]

Leave a Reply