This past weekend I attended Data Insight and hacked alongside a great group of people. Our team had a lot of fun and learned a lot by tackling a tough issue, child abuse. We used the statistics for 2002 from infochimps to visualize reported child abuse cases in the US. We wanted to do more than just show proportions by category and with some creative influence from excellent speakers we came up with this:
Check the original full width standalone version
You can mouse over the circles in the top left to view each category of abuse and see the warning signs. We hope that interacting with the visualization allows people to spend a little more time exposed to something we would otherwise rather not think about. Awareness is a large part of prevention, and we hope this can help.
Additionally we’d like to point out the lack of good data in this area. The reports available offer no insight into categories which would be useful to explore, such as breakdown by age, gender, race, family size or any other metric. There seem to be proprietary data sets which require actual mail to obtain. I particularly hope that events like data insight help show the importance and usefulness of good data. Check out the winning teams for some awesome examples
As always I’ve gotta get technical, so I want to say a few things about how this was made. It’s all processing.js with a nice Tween library from Quasipartikel. I implemented a 2D flocking particle system based on my colleague Myrna’s work. The “kids” each represent 1% of child abuse cases, so each category has however many kid particles as percent of cases. They react to the mouse with avoidance or goal oriented behavior depending on a variable set per category. They also have flocking rules which vary slightly by category to get for example more aggressive behavior or more spread out. All the code is on github so you can check it out. Keep in mind it was hacked together in one weekend!