Avoiding the Horrors of Scary Visualizations A Brief

Avoiding the Horrors of Scary Visualizations A Brief Introduction to Data Visualization October 2016

Who am I ? Jeff Pries • Been with Site. One on the Reporting team since January. • Have worked with SQL Server with a focus on Business Intelligence and creating dashboards and useful data visualizations since 2009. • 14 years in IT so far and counting! • While I’ve never set foot in the state of Michigan, I did at least grow up in another place bordering Canada – Upstate New York. jpries@siteone. com

What to Expect What this session is? A (hopefully) fun (or at least interesting) intro to some data visualization concepts and best practices. What this session is not? Technical. . . I won't actually show you HOW to do anything useful. …Also, we won’t be touching much on dashboards. They’re a whole wonderful topic all on their own.

Agenda • Tufte and Few – the parents of modern visualization • The Basics – Core Concepts • Chart Types • Color • How to Lie • Scary Visualizations

Edward Tufte (the Artist) edwardtufte. com • Started lecturing on statistical graphics in the mid 70’s. • Wrote “The Visual Display of Quantitative Information” in 1983. • Credited with popularizing / naming the Small Multiple and Sparkline chart types.

Edward Tufte • Coined the term "chartjunk" to refer to useless, non-informative, or information-obscuring elements of quantitative information displays. • Uses the term "data-ink ratio" to argue against using excessive decoration in visual displays of quantitative information.

Stephen Few (the Engineer) perceptualedge. com • Less widely known than Tufte, but takes many of Tufte’s concepts and creates practical applications. • Very “Business” and “Business Dashboard” focused. Gives the “how. ” • Credited with creating the “Bullet Chart, ” which is similar to a thermometer.

Stephen Few • Wrote Show Me the Numbers in 2004. • “Information Dashboard Design” is a one stop shop for best practices on designing useful business dashboards – everything from how to layout information to the colors to use to the types of visualizations to use.

Why do we Visualize? • The ultimate goal is to communicate information clearly and efficiently so it can be used in the most effective way possible. • Generally you want to show the data in a coherent way to get the user to think about the substance without distorting what the data has to say (whether malicious or not).

The Basics – Core Concepts Two primary types of information displays: • Tables - Primarily used to look up specific values • Charts/Graphs - Primarily used to show relationships among data (trends, etc. )

The Basics – Core Concepts Lots and lots of different types of charts/graphs… http: //www. visual-literacy. org/periodic_table. html

The Basics – Core Concepts • However, in the great sea of charts, there are only a handful -- the bread and butter, that really matter. Three of these are: Bar Chart Line Chart Pie Chart

The Bar Chart • Can be either horizontal bars or vertical bars • Great for displaying measures that are associated with discrete items in a category (such as regions or departments) • Good for emphasizing individual values and comparing to one another William Playfair, 1781

The Bar Chart - Tips Start the Y-axis at 0 Space appropriately (about half a bar width between bars) Use a single color for bars in the same series unless highlighting something specifc

The Line Chart • Best used to track changes over time, using equal intervals of time between each data point (e. g. , monthly earnings) • Good for showing the overall shape of the distribution • Excellent for tracking multiple data sets on the same chart to see any correlation in trends

The Line Chart - Tips Include 0 on the Y-axis (if possible) Label lines directly on the chart (if software allows) Plot all data points so that the line chart takes up approximately twothirds of the y-axis' total scale

Pie Chart Tips • …DON’T! …many would say that the only good pie is one you can eat. “The only thing worse than a pie chart is several of them” – Edward Tufte

The Pie Chart – Why the Hate? • Pie charts are a valid way to compare the parts to the whole. • However, they are very frequently misused. • Common Pie Crimes Against Humanity include: • Visualizing too many categories within the chart (try to stick to 3, do not pass 5) • Including percentage labels, but not totaling to 100% • Having multiple slices which are very tiny (and difficult to make out) • 3 -D

The Pie Chart • They’re not terrible for getting rough estimates, or for saying two slices combined are a majority. • Still…it’s generally better to try to use a bar chart

Color • Colors should be used sparingly and only when they bring value to the message (such as highlighting or differentiating a value). • Be consistent – a color used for a category in a chart should be consistent when multiple charts are used. • Be very careful using Red, Yellow, or Green (especially together) as they tend to indicate Bad, Warning, and Good, even if not intended to be so. • Be conscious of colorblindness. ~10% of males and ~1% of females suffer from some form of colorblindness.

Lies…filthy lies The Y-axis has been manipulated (not starting at 0) to make the difference between the two seem very large.

Lies…filthy lies The Y-axis has been manipulated (not starting at 0) to make the difference between the two seem very large.

Lies…filthy lies The Y-axis has been inverted to make like gun deaths are increasing, when they’re actually decreasing

Lies…filthy lies The cumulative graph makes the data appear as if things are going well, but the annual graph paints a different picture.

Lies…filthy lies Ignoring the black and white, comparing two pie charts is hard – but making one bigger gives the impression that those slices are bigger / more important than the smaller pie.

Scary Visualizations Unless you’re at the movies, 3 -D is not your friend…ever.

Scary Visualizations So, 13% is greater than 34% ?

Scary Visualizations Source: CNN - A donut chart is just an updated pie chart…and should still equal 100%

Scary Visualizations Way way too many categories for a pie chart!

Scary Visualizations Y-Axis mislabeled, the legend is a mess of types and colors which do not match, the background gradient is distracting.

Scary Visualizations There’s a whole bunch going on here in this Norrington visualization

Recap • Visualization is FUN, but do so responsibly. • Friends don’t let friends use pie charts.

Further Resources • "How to Design Bar Charts" - https: //visage. co/data-visualization-101 -bar-charts/ • "How to Design Line Charts" - https: //visage. co/data-visualization-101 -line-charts/ • “Practical Rules for Using Color in Charts” http: //www. perceptualedge. com/articles/visual_business_intelligence/rules_for_using_color. pdf • “Choosing Colors for Data Visualization” - https: //www. perceptualedge. com/articles/beye/choosing_colors. pdf • Reddit - Data is Beautiful - https: //www. reddit. com/r/dataisbeautiful • Reddit - Data is Ugly - https: //www. reddit. com/r/dataisugly • “Information Dashboard Design” - https: //www. amazon. com/Information-Dashboard-Design-At--Glance/dp/1938377001/ • Stephen Few’s Design Examples - http: //perceptualedge. com/examples. php • Color Brewer - http: //colorbrewer 2. org