A few weeks ago Jody Avirgan had the duo behind Dear Data on an episode of What’s The Point. During the episode he challenged listeners to create a data visualization related to each’s podcast listening habits on a postcard and mail it in. (Podcasts really love voting and postcards)
I subscribe to a ton of podcasts (well over 100), and listen as much as I am able, but I couldn’t think of a great data source.
Then I had the idea of looking at the podcast artwork of some of my favorite podcasts and networks. I first looked into using something like
PIL to create histograms for the artwork, but it wasn’t easy to read and understand. Then I remembered some frequency-type graphs of movie poster artwork and found k-means color clustering. That was what I wanted to do, but wanted to reuse someone else’s library rather than writing it up myself. After some extensive “research” (Googling), I found http://mkweb.bcgsc.ca/color-summarizer/?examples by Martin Krzywinski.
It does exactly what I want. Now I all had to do is select my dataset and translate the data visualizations to a postcard.
Here’s my postcard hitting the postmaster today.