Hans Rosling's TED talks are the stuff of legend for we data animators - his original talk, now labelled "Hans Rosling shows the best stats you're ever seen", simply stunned us when we first saw it... and then prompted us to call Visokio and persuade them to build animation features into Omniscope.

Four years ago, Rosling sold the software that powers his incredible presentations to Google - and the result was the Google Public Data Explorer. Now Google has opened up the Public Data Explorer so that you can explore your own datasets with ease... although to do so, you need to organise your data in Google's new XML dialect, the Public Data Set Language ("PDSL") - so it's one for you data pioneers only, at this stage.

Here at Atheon Analytics, having a pioneering nature, we are starting to experiment with PDSL and will post again when we have some noteworthy results... early work suggests that the Public Data Explorer remains best-suited to time-series data sets, but we have a few ideas which may work equally well. In the meantime, we are keen to hear from anyone with positive results... do let us know.

Posted
AuthorGuy Cuthbert

Data visualisation is emerging as a hot (or cool?!) tech topic - it's starting to gain momentum and demand attention from the mainstream press, vendors, consultants etc. I just watched a great TED talk on the subject by David McCandless - his choice of data sets (global spending, facebook status etc.), and the metaphors he uses to describe the benefits and potential of visualisation really help to convey the opportunity that effective visualisation yields when dealing with seemingly overwhelming volumes of data.

Yet I believe data visualisation alone is not enough.

Great pictures certainly help, and David's bespoke infographics are very effective at conveying his message, but they face two problems:

  1. Custom infographics require specialised skills and a lot of time and effort - even simple images like the 3 triangles above don't just roll off the pen or screen (if they are accurate)
  2. Static images introduce the reader to a subject, perhaps inform the reader of a few key facts, but can't convey a detailed story; every engaged reader will be intrigued by the image above, but want to know more... a good visualisation will raise more questions than it answers

The first point is best addressed by generic data visualisation applications which, whilst they can't produce truly bespoke infographics, can - and do - provide a wealth of customisable visualisations which reduce both the graphic design / artistic skill set required, and the time and effort expended, in creating powerful visualisations - three of the best such applications are listed here on this website.

To deal with the second point, however, we need to look beyond visualisation alone and consider how interaction allows the reader to explore data landscapes and uncover surprising insights.

Interacting with data forces us to consider software applications again - we can't interact with a complex infographic if it was created by hand (whether by pen or graphics tablet). To interact with data effectively we need a rapid response - ideally instant - so that we can alter the visualisation according to our investigation of the data.

The best software applications allow us to move between different 'layers' of data (from macro to micro, and back again), different subsets of data (between countries, categories, time periods etc.) using multiple visualisations at the same time to allow the reader to examine the data from several angles at once.

Such interactivity encourages exploration - 'wandering' through a data landscape, admiring views, following paths, considering options and engaging with the information present. Exploring data in this way leads to an appreciation of its complexity, unlocks hidden patterns which testify to subtleties of similarity and difference, and leads the explorer to understanding and insight - the ultimate goals of all 'business intelligence' environments.

So I propose that it is interactive visualisation that offers the greatest potential benefit when attempting to understand and act upon our ever-increasing volume of information.

Planes or Volcano? What's emitting the most CO2?

Planes or Volcano? What's emitting the most CO2?

Posted
AuthorGuy Cuthbert
"... it's what we do with that information that matters... "

In our capacity as data animators we often need to explain to our clients that what you see ISN'T always what you get, and that information isn't neutral; the meaning that you derive from it - and hence the actions that you take - depend on how that information is presented, and how you interact with, and explore, the information.

This is often hard for our audience to understand and, in some cases, they disagree with us - sometimes rigorously! Our challenge at times, therefore, is to convince an understandably skeptical audience that data and information are massively affected by context.

As such, we are always seeking new ways to explain the capabilities - and limitations - of human visual perception. This is a fast-evolving field, inspired by many disciplines including psychology, neuroscience and computer science, notably robotics and artificial intelligence.

The TED talk below (TED talks being a favourite source of ours for stretching our thinking, and dipping our animator's toes in the waters of unknown subjects) illustrates beautifully how visual perception is utterly context-sensitive. Watch out for the headline of this blog, arriving at 4:28...

Even the most hardened skeptic has to agree that Beau Lotto presents plenty of evidence for both the sophistication, and frailty, of visual perception.

Now, perhaps it would be a good idea to remove the gradient background of that latest dashboard report?

Posted
AuthorGuy Cuthbert
CategoriesVisualisation