Share of jobs in manufacturing by commuter area from 1990 to the 2016. Hover to isolate an individual commuter area’s series. Data are not seasonally-adjusted.
To keep the size of this graphic down I shrunk each faceted chart based on it’s max y-value while keeping the y-scale consistent across charts. At first I just let the browser float each chart in a container but this left weird gaps between charts. Looked into it and found that fitting irregular shapes into other shapes neatly is called the bin packing problem. I implemented a 2D rectangular bin packing algorithm in JavaScript based on Jukka Jylanki’s C++ implementation of the algorithm. This let me lay the charts out snugly.
This bin packing implementation can be found in this GitHub repository: BinPack. The API is still a bit rough at the moment.
Data come from the Bureau of Labor Statistics. The
download-data.sh shell script in this gist repository shows how to download
the data in bulk. The clean-data.R R script shows how the data was prepped
for the visualization. The little state icons are from ProPublica’s
StateFace project.