Seoul Selfie

CODE: GitHub
DATATYPE: images
SOURCE: Instagram
PLACE: Seoul
YEARS: 2014-2015
SIZE: 10,000,000 slices
VIZTYPE: flat slice histogram
FEATURES: hue, brightness
FEAT SRC: scikit-image
MORE INFO: Software Studies Blog

This is a "flat slice histogram", a slice histogram where each bin is forced to a fixed height. Normally, histogram bins are fixed "width", meaning they cover the same range of the distribution variable. Here, the bins are fixed height, and thus the variable ranges they cover are unknown (without additional markings on the plot, which is of course a possible modification here). Note that the flat histogram would be perfectly useless for traditional statistical plotting, but since we are making image histograms, they can reveal a lot. The left-to-right visual patterns do still map roughly to the distribution variable. Here, hue is the distribution variable, and we sort vertically by brightness (a couple times). We can see quite clearly that in this dataset, blues and reds tend to be darker than the greens and yellows.