Mr Andy Kirk, Founder of Visualising Data joined us yesterday at Big Data World Europe 2012 and presented on Understanding learning in order to implement efficient visualisation methods. He touched upon:
- Examining what data visualisation is attempting to achieve
- Assessing how lessons learnt from psychology can be applied to improve visualisation methods
- Understanding how visualisation can provide a more balanced assessment of data, overcoming the natural bias that collecting your own data incurs
- Building visualisation techniques upon a psychological framework to ensure they work better with human mental representation and visual processes
Guest blogger Paul Booth shares his views, tell us what you think below!
If you don't know Andy Kirk's website and you are interesting in how data should look, take the time to check out http://www.visualisingdata.com/. There are plenty more blogs and sites focused on the visualising efforts of those working with large and not so large data sets. During the talk Kirk talks refers to a skills gap, with little or training being given in design for those who have to present and explore data. To some (and I have witnessed this) design is just about making things â€˜pretty' or choosing better colours. But thanks to a lot of work in the field on visual perception, and books from Stephen Few (Information Dashboard Design), Jacques Bertin (The Semiology of Graphics), Ben Fry (Visualizing Data) and a few others of note. Kirk offers advice and insights into principles vs heuristics – there are no hard and fast rules, but its not about being touchy-feely either. We are taken through shocking examples of corporate dashboards, how exploring visual perception abilities have increased the understanding of maximising data comprehension through the human eye and into the brain along the path of least resistance. Kirk also talk about cognitive science – known as gestalt theory which is around 100 years old now, but brings fundamental insights about the perception of shape, pattern and visual deceptions in size, colour and depth.
A few very helpful links appeared in the talk such as colour brewer which can help people get an effective colour system for maps. More specific than this was the well informed idea of visual variables- for quantitative and qualitative data applying position as a visual variable (an attribute for a shape or symbol) is the most efficient way to display data. This is because humans can distinguish between vertical and horizontal positions in a 2D space better than size, shape, colour or any other. Think about, it makes perfect sense, and there is a handful of studies to support this.
To wrap up the talk Kirk talks about DV as a craft – not a science and outlines his own â€˜8 hats of data visualisation' – illustrating the multidisciplinary nature of working with visualisations. A valuable talk indeed, but I would go a little further and push employers to take data visualization as seriously as â€˜data science', or analysis. It isn't the qualitative and â€˜woolly' area of design that management can often think it is, the problem comes when decision makers have no understanding and no design vocabulary themselves, how do know what they need or what can be delivered with that? Following the other talks asking â€˜does your company need big data now?' and other technical aspects, I hope people realise the next set of questions will be â€˜what does my data look like?', â€˜how can I see if there are issues?', â€˜how should I show my data to my customers, shareholders and staff?'. Check out Andy Kirks blog (the slides will be up from the talk soon) and get into design in a big way, you'll need it.
Presentations from the event will be available from the 24th September!