In official statistics we’re used to dealing with highly aggregated data. To visualise those, bar-, line- and pie charts are standard tools. But there is a whole other side to visualisation where it is used to recognize patterns, outliers or errors in individual data.
A very compelling example of displaying daily data for several years in a calendar view was recently presented at SVGopen 2011 by Mike Bostock, a Tufte influenced data visualisation expert, who could easily become the next Hans Rosling. Mike is among other things author of the d3.js library, a JavaScript Toolkit that can streamline your data visualisation projects a lot without forcing any constraints of predefined chart types on the developer.
In the above excerpt of the daily flight data for 1995–2008 (click image to see all the years covered) the effect of 9-11 is the most obvious, where flight activity stopped completely and had a hard time to recover for several months.
But there is so much more information in there:
- the weekly pattern: less flights on weekends
- the seasonal pattern: most flights in July and August
- the 9-11 pattern: few flights on 9-11-2002
You might want to flick through Mike Bostock’s SVGopen 2011 slide deck to get an idea what else to expect (use arrow keys to navigate the web based presentation). All this is done using open webstandards of the latest incarnation, i.e. HTML5, CSS, JavaScript and SVG, which means there are no plugins required.
You can watch Mike Bostock live at W3Conf: Practical Standards for Web Professionals on November 16th, 2011 or later as a video download.
Blog about Stats 