This is a modified scatter plot I dreamt up to display more than two variables without using bubbles. It’s not a finished product; just an idea that seemed like it might have some potential. I built a simple test version in d3 and would welcome feedback from the community.
Here’s the link to the interactive version
https://beta.observablehq.com/@tippytappy/scatter-plot
You can paste your own data as a csv to plot it.
Looks interesting, Dylan. Reminds me of the OECD chart by Moritz Stephaner, http://www.oecdbetterlifeindex.org/, but your thin lines are more suited for dense displays than the stylized flower petals.
Seems most useful when there are two important variables for the X and Y axes and some secondary variables for the radial needles, since the dot position will be most accurate. I tried the Iris data with the first two principal components as X and Y and the four measurements as the radial needles. Not sure it’s that useful, so maybe it’s best to keep the number of secondary variables small as you have, at least when the points are relatively dense.
Being able to represent negative values is certainly an advantage over bubble sizes or glyph/radar charts.
Attached image:
Thanks, Xan. I wasn’t aware of Moritz Stephaner’s chart. As a fan of his work it’s good to think I had a similar idea. I agree that this chart suits some situations better than others. This was one of those ‘scratch the itch’ projects; I had to see it in action to know whether I was on to something.
I’m a big fan of your work too. I first saw JMP when I came to the SAS campus for Jorge Camoes’ talk. JMP is brilliant. As an R and Tableau user it felt like the best of both worlds in one package.
Thank your for your feedback.