[infosthetics@strataconf 2011 by guest blogger Collin Sullivan] Greetings from sunny Santa Clara and the O'Reilly 2011 Strata Conference! I have been prepping for what are sure to be enlightening and challenging presentations here, and am looking forward to probing great minds. The conference is off to a fantastic start, with the day's long tutorials already under way.
This morning I caught Drew Conway's opening presentation at the Data Bootcamp session, an introductory tutorial on the practice of data science. It was not long before it had become clear that I was in a bit over my head -- the explanations were clear but the content was very programming-heavy, focusing much on the nuts and bolts of packages like Python and R. Of course there was much talk of information design, though, and Drew's 3 rules for data visualization provided me a sure base from which to operate: 1) Make complex ideas simple (simplify); 2) Present information from lots of data in compact and consumable sizes (parsimony); and 3) tell the truth (honesty).
I spent the rest of the morning a few doors down at a session that expounded upon these bedrock principles. Zach Gemignani, of Juice Analytics, discussed the foundations of dashboard design during the Interface Session titled "Making People Fall in Love with your Data". He challenged the crowd to combine the defining characteristics of the salesman, the therapist and the connoisseur in their approach to dashboard design. That is, a designer should bring a salesman's motivation and knowledge of the user's desires, a therapist's ability to identify hidden problems and empathize with the user, and a connoisseur's reluctance to settle for anything less than the very best.
Gemignani's approach is user-centric: to him, all that matters is the experience the user has when interacting with the data. He outlined lots of rules and ideas -- the dashboard ought to guide the user through her journey, the focus ought to be on actions that can be executed--mainly by employing a house design allegory, but all focused almost exclusively on improving the user's experience. (And note that you can download his presentation slides here.)
For those seeking examples of the well-done and the beautiful, this session did not disappoint. Gemignani ran down a list of some different types of dashboards that were both visually appealing and effective at conveying data efficiently. He showed us:
The Leaderboard. Quickly answers a common question about the best and worst in a data set.
The Drill Down Workflow. A visually intuitive approach in which the user's selections, made from the top down, alters the dynamic output.
Simple graphs. This one demonstrated that data can be represented in a striking and effective manner, without being overly involved or complicated.
Actionable graphs. These infographics give the user something she can use immediately. Also note the dynamic text summary ("Buying is better than renting after 6 years.") is prominent, clear and concise.
And what might be called the All-in-One. Customized with user input, creatively organized, guiding the user from beginning to end. And a beautiful presentation, at that.
Gemignani's presentation served as a useful primer for those unfamiliar with dashboard design approaches. It was not couched in jargon, nor was it overly technical. Indeed, Gemignani focused on the end user--his audience--and conveyed his information in an efficient and clear manner. Much like a salesman. Or a therapist.
And with the way he took apart that Federal IT Dashboard (see the slides linked above) and put them back together, he is definitely a connoisseur.
More in the next 2 days!
This post was written by Collin Sullivan. He is a research analyst for The Sentinel Project for Genocide Prevention, where data collection, analysis and visualization are being used to design an Early Warning System (EWS) to detect and prevent genocide. Collin lives in San Francisco. You can reach him at collin [at] thesentinelproject [dot] org and follow him on Twitter at @inciteinsight.
StrataConf 2011: Making People Fall in Love with your Data - Now let's work on astronomical data!
Posted by Alberto Conti at 15:05