It seems visualisation of data is a current hot topic, most days I get a suggestion that a certain data set might be better presented in a visual format, news sites such as the Guardian and BBC seldom run an article without one and I have really been enjoying the CETIS OER Visualisation project. What is it about them that makes visualisation so attention grabbing?
Yesterday Mark made the comment on his blog that while pretty the twitter visualisation graphs may be useless and offered some insight the role that he thought Visualisation techniques play. While I agree that the twitter graphs are useless; my view on why visualisations may be a powerful tool seems to come from a different place to marks.
Recently I have become interested in patterns, not so much the machine learning pattern recognition (although that is interesting!) but also recurring patterns in history and society. This (as usual) has stemmed from one of my more geeky interests. In the 1940s sci-fi author Issac Asimov started writing a series of short stories that would later turn into a book entitled Foundation. Before writing these short stories Asimov had read “the History of the decline and fall of the Roman Empire” and spotted patterns in history. Based on these patterns he started to write short stories exploring methods that societies use to progress and adapt to change.
The stories revolve around the fall and rise of a future galactic empire and reading the book today is quite chilling as the book talks about a fictional method of analysing sociality and making predictions as to its future which sounds very familiar today, 70 years later.
I think the reason why I am so attracted to the Foundation stories is because Asimov has identified recurring patterns in history and using the format of science fiction stories is able to explain why these patterns are relevant to the context that we live in today.
Patterns in Data
When I think of patterns I think visually and for this reason I think that visualisations are the perfect way to explain patterns we have found in data to each other.
The problem I have with many techniques and the reason that I find pretty twitter graphs useless is their ‘stillness’. I guess that patterns are usually found either in time or between data sets, something that I don’t see in a bunch of lines connected to each other.
Mark suggests that visualisation is a performance art and requires a story to be told along with the visualisation. I think this is probably true in some respect, but I guess I would like to see better visualisation techniques that can express how an author sees patterns in the world and these patterns emerge. I’m sure there is a cross over here with upcoming modeling techniques such as agent based modeling and statistical pattern recognition.