We were asked, recently, to present to the global sales, marketing and supply chain presidents of one of our customers on what the UK business had achieved using visual analytics. This was a great opportunity, but preparing for the meeting led to an interesting question:

"What are the key attributes of the visual analytics applications we use which makes them great for visual analytics, which other software tools don't posess?"

On the face of it, this should be easy - they all offer great visualisation, of course... but this isn't unique to them (there are a great many visualisations available across a multitlude of tools). As we considered the question further, we realised that our approach to visual analytics - the Atheon way, if you like - is dependent equally on three key attributes.

1. Visualisation

Yes, visualisation comes first in the list. Visual analytics without visualisation is impossible, so we need to be able to present information in a visual manner, helping our audience to see the patterns in their data rather than try to calculate them.

Most of the people we work with, the audience for the work we do and the applications we build, are business people - very few are full-time analysts, and few have any substantial formal training in statistics. 

The tools we use - and the applications we build with them - start with visuals, and enable the user to explore those visuals before attempting to quantify precise values through data tables. Traditional business intelligence tools - spreadsheets included - start with tables of data and allow the user to create a visual representation here and there.

2. Interactivity


When an enquiring mind is presented with a clear visualisation, an improved understanding emerges... but the enuqiring mind immediately conceives another question which itself can be answered by a different, or altered, visual. Our aim is to serve up image after image to allow the user to explore their data as quickly as they think of questions; this high degree of interactivity - almost real-time in many cases - encourages deep engagement with the data and draws the user into a richer understanding.

Traditional BI tools do allow interactivity, but this can often mean slow round-trips to huge database servers, resulting in tens of seconds - even minutes - between questions. This slow response actively discourages the user from exploring the data, and leads to a limited understanding based on pre-canned reports. 

3. Data blending

Data Blending

Rapid interactive graphics enable the user to explore data at speed... and this leads to the third essential feature of the tools and techniques we use; data blending - the ability to stitch together different data sets, from different sources, in order to build the most complete and robust picture possible.

Despite the best efforts of IT departments around the world, much of the data that organisational users depend on resides in departmental spreadsheets, files received ad hoc from customers and suppliers, or drawn from an increasingly large range of sources on the Internet. It is impossible for IT to own and manage all of these data sources, and so traditional tools simply ignore them... this explains why so many expensive BI tools lie dormant, with data extracted in Excel and 'blended' through the high-risk vaguaries of VLOOKUP.

The tools we use can all access multiple data sources at once, encouraging the user to merge different data sets to create rich combinations that provide the best view of the problem, or opportunity at hand.

For more about the tools we are talking about, take a look at our product comparison.

AuthorGuy Cuthbert