The world of data visualization maintains as tight a canon as any field. Float around for a few months, and you'll see Minard's Napoleonic war visualization and Nightingale's Coxcombs or the 1870 US census atlas dozens of times. For new data, there's nothing like the Social Security Administration's database of names. People love them some names. (If you've never seen any of these: One of the more interesting ones I've seen lately is David Mimno's topic model of names treating each state-year combination as a topic)
So this seems like a decent way to test out some elements of the bookworm library for text
analysis. Mimno's text-document model is a nice way to dump the
Social Security data into tools designed originally for
bag-of-words text analysis. So that's what I've done here: treat
the name data as 10,000 books (one for each state-gender-year
combination over the last 100 years) with the names of every child
born listed. From there, I've basically just dropped it into the
Bookworm system. Consequently, there are going to be some weird
things going on here. It says you're searching for "words", but
really you're searching for names. Straightforward line charts are
available here. This one
here uses a heatmp, which I've found to be the most useful way of
exploring bidimensional data (here, set up as year and state).
The really interesting part of this for me had nothing to do with the
data itself: it was about how
to put the states in order. Read that link for an excessively
elaborate description.
Oh, and click on that colorbar to switch axis types. It's awesome, I promise.
Made by Ben Schmidt. See more maps and visualizations