Data visualizations | Joe Mitchell-Nelson

Data visualizations

Below are some “extra-curricular” data vizualizations I’m especially proud of.

I originally made the app below to help my wife think about where we might move after grad school. It’s a strictly unscientific tool that’s meant to be played with, to get a sense of where you might like to live. Move the sliders right or left to indicate the importance of each characteristic to you and receive a list of the counties that maximize those characteristics. You can also impose hard cut-offs for some variables. Behind the scenes, each slider controls a parameter in a linear utility function. The variables are transformed to be z-scores, to give them (somewhat) comparable units. Code/data are on my Github. You’ll need an API key from the Census Bureau to run the code, which can be obtained here.

Notes: Climate security uses county estimates from Hsiang, Solomon, et al. (2017). High school graduation rate data comes from, and partially imputed using demographic information from the Census Bureau. Air quality data also comes from also provides incomplete violent crime data, which I supplemented with data from the FBI's Uniform Crime Reporting. Mild summers and mild winters calculate the highest average monthly maximum temperature in the hottest month (for summers) and lowest average minimum temperature (for winters), which are then each averaged over the last decade in each county. Data comes from NOAA. Distances to MSAs are calculated as distance from county-centroid to MSA-centroid using shape files provided by the Census Bureau. Politics is determined by 2016 vote shares in each county, as reported by the MIT Election Lab. Housing affordability is measured as median monthly housing cost, as reported by the Census Bureau's 5-year American Community Survey. My wife and I (separately) consistently receive recommendations for counties in Oregon, Washington, and Colorado. Sometimes we also get counties in the Great Lakes area or the suburbs of D.C.

Some recent studies have found significant effects of COVID-19 lockdowns on air quality, especially in cities like Delhi, Rio de Janeiro, and Wuhan. I was curious to see if similar effects were apparent in urban areas in the U.S., so I plotted 2020’s PM2.5 data for New York City, compared to recent years. The code and data to produce this plot are on my Github.

I handled technical implementation to produce the animation below, in a joint effort with Joanne Sherrod. It was featured in Fast Company, tweeted by AFT president Randi Weingarten, and shared on Facebook more than 7,000 times. You can find the code on my Github.

"I don’t take responsibility at all"