Launch, grow, and unlock your career in data

June 19, 2017

Makeover Monday: Is America Improving Its Ozone Air Quality?

Eva and I are hosting another BrightTalk webinar this Friday titled "Visualizing Smart Cities: How Data can help shape our communities". This is part of an Internet of Things week for BrightTalk so when deciding what data set to pick for this week, I got to thinking about air quality sensors. A quick google search turned up a treasure trove of data on the EPA website. Thanks for my friends over at Exasol, I was able to create a place for all of us to play with all 202 million hourly measurements. If you want to play with the live connection to all 202M records, you will need a login which you can create via this registration if you do not already have one.

The EPA website also contained some basic reporting, which is the focus for this week's makeover.

What works well?

  • Simple layout. It's easy to see you have days within months going left to right and years going down.
  • Colors match the official EPA colors for the AQI categories
  • Including the ranges we know what constitutes a good reading vs. a bad reading
  • The title tells me what I'm looking. Including the date range is a nice touch too.
  • Using a strip plot along with the colors helps reveal seasonal patterns
  • Including the source in the footer

What could be improved?

  • Make the colors color-blind friendly
  • It's difficult to see how a month has changed across the years.
  • I can't hover to see the values.
  • It says each “tile” represents one day of the year and is color-coded based on the AQI level for that day, but I think it's actually showing the max measurement for each day. It should be more clear what they are representing.
  • There's very little context. How does one county compare to others? How does one county compare to the national average? 

What were my goals?

  • Last year I challenged The Data Duo to visualise an almost identical data set. Pooja created a pretty amazing viz (of course), so I wanted to pick some parts off of it. Particularly, the State selectors on each site and the dot plot.
  • I wanted to understand how each month changed across the years, which is pretty much exactly what a cycle plot was created for. I used a trend line instead of an average though because that seemed to show the patterns better.
  • I wanted to add context for comparisons to the national average and to the state average (which appears when you click on a state).
  • I wanted to use color-blind friendly colors.
  • I wanted a small sparkline-type chart to show how things have changed since 1990.

With these goals in mind, here is my Makeover Monday week 25 submission. Click on the image for the interactive version.


  1. Very interesting dataset, thank you. I have been trying to reproduce the original datavis, but I do not get the same numbers using the datasource from your tableau file. However, if I use the other one linked from the makeovermonday website, I get the same numbers if I use "max" instead of "avg" for the AQI, which I think is correct (since AQI uses maximum value for a given day).