People say that a picture is worth a thousand words. The human brain and visual system is absolutely amazing in how it can take a picture and turn it into meaning. An important part of any analytics platform is to create pictures that people can interpret and make decisions from. What are some basic principles of good visualization?
Show the data
One cannot talk about principles of visualization without mentioning Edward Tufte. In his book "The Visual Display of Quantitative Information", his first principle of good visualization is to show the data and show as much of it as possible given the constraints of the screen or paper. It's important to remove unnecessary decoration and focus on the data. One might think that showing too much data can be overwhelming, but according to Tufte, "There is no such thing as information overload, just bad design. If something is cluttered and/or confusing, fix your design." Tufte has some great comments on how well the IPhone follows this principle.
Color
In the book Envisioning Information, Tufte said “… avoiding catastrophe becomes the first principle in bringing color to information: Above all, do no harm.“ Lots of bright loud colors can be unpleasant and distracting. Certain combinations of colors can cause strange visual effects. Maureen Stone has a great blog article about this. Color blindness is also an issue that needs to be taken into account. Ten percent of men are red-green color blind. If Sports Illustrated showed for example a line chart with a red line and a green line, ten percent of their audience wouldn't know which is which. Maureen Stone again has more to say on the subject of color blindness.
Interactivity
A good visualization should be interactive. If you have ever heard a talk by Ben Schneiderman, an expert in the area of Information Visualization, you've probably heard him say, "Overview first, zoom and filter, then details-on-demand". Allowing users to get a high level view of their data and then zoom in for more details can be very powerful. A simple example of this shows the most popular baby names in the United States since 1880.
Animation
Hans Rosling has shown that if done well, animating data can help tell a story and make the data more interesting. Animation can also help in transitioning between different views of the data as it can provide context that would otherwise be lost. Of course animation when done badly can be distracting and annoying as shown by many of the banner ads on the internet. There are also few things more annoying than waiting for an animation to finish when you are trying to get something done. Animations should generally be quick, subtle, and have a purpose.
Beautiful
Visualization can be beautiful when done well and can be art in and of itself. For more visual eye candy check out some of my favorite blogs:
- Infosthetics
- Visual Complexity
- Flowing data
Pattern Based Analytics is an emerging area of advanced analytics focused on simplicity and transparency about conclusions. As more and more data is collected and rapid response becomes the norm, quickly finding patterns and insights about future outcomes becomes more important. Ultimately, this means more proactivity, less reactivity.
See our LinkedIn Group to join discussions about Pattern Based Analytics.
A challenge that has always been present with data and will continue to be a challenge with pattern discovery is "what does the data mean" and "how does the data I'm using impact my analysis"? Visualizations have been facing these challenges for a long time. The significance of patterns depend on the observer (which typically involves a visualization) - the data or the system generating the data doesn't care what it looks like.
ReplyDeleteFor those of us that are merely Computer Users and not at all interested in what is going on inside our machines, visualization is critical to getting as much as we can (or need) out of our computers. Over the years, I have switched from being a PC user to a Mac user then back to PC – and then from the old Microsoft (2003) with which I was comfortable to the new Microsoft (2007) where they changed the entire toolbar thing with the buttons. And each time, I went through a difficult transition period – I’m still transitioning to this new Microsoft and it’s been almost a year. It seems to me, when designing new features for a software product, the last thing you want to mess with is the visualization.
ReplyDeleteIt occurs to me also, how huge the challenge is when designing a new software platform. You have to get it right enough that you aren’t changing it dramatically with each new release but you can’t possibly get it right enough without having ample input from new users – it’s a Catch-22. I find myself with a new-found respect for the masterminds developing visual displays for the mere Computer Users of the world.