blog fossil diary
I just realized that I posted this to Twitter and not anywhere else; whoops!
Anyhow, I’m an avid reader of FlowingData, because Nathan Yau, the man behind it, does some pretty awesome stuff. Â His visualizations are clear and still aesthetically pleasing, and his concepts are always nice. Â Of particular interest to me, when I first started reading, was your.flowingdata which is a means to track your own life through Twitter - for example, you can tell it when and how far you ride your bike every day and have it automatically generate a visualization of distances ridden over time.
Recently, however, he posted a little challenge of sorts. Â Given a dataset, we, the readers, were to visualize it our own way and draw some conclusions from our visualizations (that, after all, being the point of visualizations). Â I’d never done anything like that before for various reasons. Â I didn’t want to learn a new domain-specific language such as R that would then require me to edit my results in the form of an image in some other program such as Gimp or Inkscape. Â Also, Gimp and Inkscape have some quirks that I’m still learning, and I didn’t want to have to chose between learning those and buying Adobe CS. Â However, I have been working quite a bit with Javascript recently, so it seemed to make sense that, when I found two libraries - Flot and Protovis - for visualization in JS that I go ahead and use one of these ‘Visualize This’ challenges to learn one of them. Â It’ll definitely be helpful in the future.
The most recent challenge was visualizing data from the National Survey of Sexual Health and Behavior. Â Given a small set of data - percentage of respondents in different age groups admitting to engaging in nine different behaviors over the past year - I worked hard to learn Protovis from scant documentation in order to pull together a visualization. Â Since it takes place over three ‘slides’ and has text to go along with it, I’ll let it speak for itself here.
I think I did fairly well, given the fact that I wound up doing exactly what I didn’t really want to - learn a new DSL. Â Granted, this one will be useful in my web design in the future! Â With the time limitation of a due date and the fact that I was learning as I went, I didn’t quite pull off exactly what I wanted, and the trends I was interested in looking after weren’t as apparent I was hoping. Â The problem was mostly due to inadequate documentation on Protovis - much of the documentation that wasn’t simply API documentation was either examples or brief write-ups about concepts in statistics as the applied to Protovis. Â I learned most from the examples, after I learned some of the basics from the API docs.
I’ll probably find another dataset somewhere that interests me in order to visualize it soon, but I also expect that I’ll be implementing the visualization process in my own projects as well. Â I’ve got lots of ideas.