Comments on: Trajectories of Obesity

By: Kyle Peterson

Kyle Peterson — Fri, 02 May 2014 14:03:42 +0000

Thank you for your kind words, Ira. The short answer is that I created a scatterplot with a “spacing” value as the Y-axis, the years as the X-axis, and the percentages as the bubble size. I did a lot of touch-up in PowerPoint to get the final product I posted. (I have a very limited tool set…)

If you send me a PM on Twitter (I’m @kpeters0n) with your email, I’d be happy to send you much more info and a clearer step-by-step instruction.

By: Ira

Ira — Fri, 02 May 2014 10:12:21 +0000

Hi Kyle. I would like to know how did you build that bubble chart? It’s amazing.

By: Jon

Jon — Sat, 05 Apr 2014 20:22:48 +0000

Another option for a static version is to use small multiples and grey out all the series except the one (or maybe two) you want readers to be able to see (see picture).

An interactive version might look like the “Interactive Time Series Chart” tutorial Nathan Yau over at Flowing Data did a few years ago: http://flowingdata.com/2012/04/04/interactive-time-series-chart-with-filters/. There, the lines are all grey until the user chooses one and that line then turns color.

By: Kyle Peterson

Kyle Peterson — Fri, 04 Apr 2014 21:43:14 +0000

The last column should, of course, be 1996–2005, instead of 2004. Sorry about that.

By: Kyle Peterson

Kyle Peterson — Fri, 04 Apr 2014 21:42:25 +0000

I think a bubble chart would be a good way to show this data in an interactive way. I’ve mocked up a possible bubble chart here. This helps compare across age groups as well as by year of birth. You could do the same for gender and race to make it interactive.

By: Joyce

Joyce — Fri, 04 Apr 2014 05:23:56 +0000

Thanks for the suggestion! It looks much better!

By: Jeff Harrison

Jeff Harrison — Fri, 04 Apr 2014 02:10:25 +0000

Another thought that occurred to me is that the lines moving to the left may not intuitively communicate the trend. I’m wondering whether this arrangement, showing profiles by age instead of birth year, might be a clearer way to encourage that comparison.

By: Jeff Harrison

Jeff Harrison — Fri, 04 Apr 2014 02:08:01 +0000

A big problem is all the cross referencing that has to be done between the data and the legend at top. There’s nothing about the colors of the lines that suggests a sequence, so even if you understand what’s being graphed it’s not obvious what the conclusion is. Here’s an approach that puts the data labels on the graph.