Chart Your Way to Visualization Success

Chart Your Way to Visualization Success

Following up on the themes of Sarah’s earlier blog, “4 Cartography Color Tips Hue Should Know”, here are a few tips I picked up from DataWeek 2014 in San Francisco in September 2014:

Visualizations and infographics are a powerful way to communicate data. However, with great power comes great responsibility, so here are a few ways to make sure they turn out clean, beautiful, and well-suited for their purpose: to be shared with the public.

Use The Cycle of Visual Analysis

Tableau guru Mike Klaczynkski defined the cycle of visual analysis as a six-step process that’s applicable across a broad range of data analysis:

  1. Define the question

  2. Get data to answer the questions

  3. Structure and clean the data

  4. Visualize and explore the data

  5. Develop insights

  6. Share the results

Simple enough, right? The last step, however, is a doozy. If you’ve gone through the trouble of steps 1-5 and then don’t share the results clearly, it could unravel all that hard work. As a data professionals you should provide legible visualizations to share your results with your intended audience.

Check Your Charts Before You Wreck Your Charts

When producing a visualization, do what Dave Fowler from Chart.io recommends and ask yourself: Am I trying to impress people with how cool this looks? Or am I trying to share my results clearly? If you’re more concerned about bells and whistles on your visualizations, you’ll end up graphics from a 1997 clipart nightmare instead of a powerful way to stream your message. Use the eight steps below and chart a voyage away from the rocky shores of bad decisions:

  1. Make your visualization audience-appropriate. You might not use the same chart to explain something to your dad as you would for your fellow data analysts. You might if he were also a data analyst.

  2. Make a graphic appropriate to the data (e.g. don’t make a time series for something with no time component). This site has a great breakdown on what kinds of charts to use for what kinds of data.

  3. Make sure it’s not a pie chart (people can understand square area better than they can circular areas). Read Death to Pie Charts to learn more and also get a bunch of great visualization tips.

  4. If you’re making a map, make sure it’s not just showing population density, as pointed out in this excellent example from webcomic XKCD (which has been linked before in a previous Atlas Blog about bicycle and pedestrian crashes in Philadelphia by Daniel McGlone). Sometimes you can get around it by normalizing the data by population.

  5. Avoid skeuomorphism in your charts, or trying to make an object look like the thing it represents. While there’s still some debate about whether websites and apps all need be stop being skeuomorphic, but there’s no question that pseudo-3D charts with photos of bananas on them need to go:

  1. Ask yourself if you’re showing the data clearly.

  2. See if someone unfamiliar with the results can interpret it.

  3. Show it off to everyone!

Chart Your Journey to Better Visualization

There are a ton of resources out there to make sure that your visualizations look good and get your message across. Get a head start by checking out the beautiful infographics blog, Information is Beautiful, thumbing through books by legendary visualization experts Edward Tufte or Stephen Few, or trying your hand at a cornucopia of data visualization tools at Datavisualization.ch:

Remember, the point of any visualization, whether it’s a chart, graph, or a map, is to communicate data to an audience in a meaningful format.