Dashboards. Infographics. Vizzes. Everyone is talking about the cool buzzwords in Business Intelligence. But, with statistics like “70-80% of business intelligence projects fail” (Gartner analyst Patrick Meehan) floating around, folks are tripping over jargon and falling on their faces. Even in a perfect world where requirements are clear, business questions are documented, and key metrics are defined, communicating effectively with data visualization is still just plain hard. Here are five tips to keep you on track and to help make your vizzes both cool AND effective.
1. Choose the Right Visualization
Talk with whoever will consume the report (have a conversation, in real life if possible – I know it’s scary, but you can do it). Make sure the visualization you choose caters to their specific needs. Do they just need that one number so they can copy and paste it into an email every Friday? Great. Then just report the number. Don’t hide that number behind colors, bars, lines, or added drama. Do they need historical context to explain that number? Fantastic. Then add a line chart with that metric over time. Do they need to compare that number to a benchmark? Wonderful. Then build a Gantt chart with the number and the benchmark. In data visualization, less is more.
2. Be Deliberate
Fancy Business Intelligence software and even good ol’ Excel love to add gridlines, borders, shading, legends, and labels to every piece of every visualization. Resist the urge! Fight back! Embrace minimalism and let your data be the center of attention instead. Use descriptive labels and legends to help communicate your message rather than clutter it. Remember, white space is your friend.
3. Color Matters
Green means go. Red means stop. Deep saturated red means you cut yourself shaving. Whether we like it or not, color carries inherent meaning. Use color to call attention to some of your data or to emphasize a point. But, be deliberate! And consistent. Does blue mean profit in one chart and loss in another? Perhaps a bad idea. Remember, colors are fun, pretty, and cool. But. if they don’t add meaning to your visualization, consider simplifying and going with a classy shade of grey.
4. Cool Or Confused? Intercepts and Axes.
Nobody loves a hacked off axis more than news networks and politicians. The drama! The intrigue! Does it inspire shock and awe? Yes. Is it misleading? Very. Avoid confusion and err toward accuracy by setting your axis intercept to 0.
Another example of “cool” that quickly turns into “confusing” is the dual axis chart. Consider how your audience will interpret your visualization. Did orange outperform blue? Are those metrics even comparable? When in doubt, break those bars and lines into separate graphs. And if you insist on keeping them in one graph, for goodness sake, clearly label those lines and axes.
5. Oh Those Pie Charts… And Maps…
Pie charts are incredibly fun to create and terribly difficult to interpret. Slices often look similar in size when they are quite different. And too many slices can set your end user up for a Where’s Waldo? game to find the right slice. Be kind to your end user. Recommend a better visualization.
Maps, like pie charts, are a ton of fun to create. The colors! I can even see the ocean! However, think twice before you slap that data on a map. Is this really a geospatial analysis? If you want to identify the state bringing in the highest profit, looking through a bunch of numbers on top of a map is just another Where’s Waldo? game. Once again, be kind to your end user. Choose a better visualization.
Now Ignore Everything I Just Said. Sort of.
Rules are made to be broken, but it’s important to learn and understand the rules before you break them. Famous painters like Pablo Picasso and Matisse studied and experimented with principals like color theory and perspective before breaking those rules with their more modern paintings. So, once you have a handle on some data visualization basics, don’t be afraid to experiment, push back on the hundredth pie chart request, and have some fun. Happy Vizzing.
Originally posted on Data Science Central, by Matthew Sundquist.
New! Plotly lets you style interactive graphs in IPython. Then, you can share your Notebook or your Plotly graph. It's like having the NYTimes graphics department inside your IPython.
Here's a preview of how it looks to have your code, data, and graph all interactively available. See the live version.
Note: this page contains paid content.
Please, subscribe to get an access.