You probably already created some Data visualizations when you learned how to use a library such as matplotlib or plotly. Maybe you even frequently use visualizations for exploratory data analysis (EDA) or dashboards. Or you are just a beginner and still in the process of figuring everything out. Wherever you are in your journey, I want to invite you to look past the code and discover the basic qualities of great data visualizations, so that you can use them as well.
"Most of us need to listen to the music to understand how beautiful it is. But often that’s how we present statistics: we just show the notes, we don’t play the music."
- Hans Rosling
According to Alberto Cairo, there are five qualities that a great visualization possesses.
- Truthfulness
- Functionality
- Beauty
- It’s insightful
- It’s enlightening
Let me explain what they mean and why they matter.
Truthfulness
A truthful visualization is honest with it’s audience, it presents data in a way that is not misleading. This not only means that you shouldn’t hide or obscure data in any way but also that your design and data choices, such as the way you created categories, need to reflect the truth and be easily interpreted by the audience. Notice that the interpretability is actually part of the second quality: Functionality, but the five qualities are not independent.
The second important part to create a truthful visualization is to avoid self-deception. Before creating a graph or visualization we need to check if our interpretation of the underlying data is correct or whether there may be alternatives. We should try our best to avoid any personal biases within our visualization, which sometimes is easier said than done. So before you start creating a visualization right after you found an interesting relationship or pattern in your data, stop and try to look at it from other angles.
Functionality
Your visualization has the goal to showcase data or certain insights in a meaningful and comprehensive way. One of the most important steps when creating a visualization is to make sure your audience understands the message and to help them interpret your chart correctly. One of the first steps towards this is choosing the right type of chart to display your data or insight. This goes beyond "Don’t use a pie-chart". You need to be clear what you want to display, patterns, trends, differences etc., only plot what is necessary to convey your insight. Once you are clear on that you can choose one or more chart types to convey your message.
This website is a great tool if you are not sure which chart type might work:
You may need to try a couple different graphic forms or even ask outsiders for help in testing which chart conveys your message best, because your audience might interpret the visualization in a way you haven’t anticipated. Be aware that data visualizations and their correct or incorrect interpretation may have very real impact.
Beauty
Now, I know you are going to say "But beauty lies in the eyes of the beholder!", and in a way that is of course true. However there are certain things that many people consider aesthetically pleasing and you should aim to incorporate those into your Visualization. The data visualization should be perceived as visually pleasing by as many people in your audience as possible, after all you want people to look at it. A good rule of thumb is to keep your visualization simple, clear and to display the data in an elegant way. Yes that means you should usually not use 3D or chart junk.
If you are still not conviced I challenge you to find a person that likes the first chart better than the second one:
It’s insightful
Although we just talked about beauty, the main goal of a data visualization is to create (new) insight for the audience. It should facilitate discoveries that would have been impossible if the data was presented in a another way, such as a table. This means that whatever you visualize should go beyond the immediately obvious. The insight your graph delivers could be an "aha"-moment, but it could also be more of a "hm I’ve never looked at it this way before, interesting!"-moment. If your data visualization at hand doesn’t do that in the slightest, it might need some rework.
It’s enlightening
This quality derives from the application of the previous ones. If you make sure to hit the first four qualities you probably created a very good visualization. There are many very good visualizations, but the truly great ones deliver on another end: they change peoples minds or lead to actions that matter. The key ingredient for this is the topic of your data visualization. Changed minds and actions only occur if you are addressing relevant issues. This doesn’t mean that every graph needs to be about pressing matters, or that visualizations about "fun stuff" are worthless, on the contrary, they are important and valuable. Yet, if you can use your skills to create a great visualization that creates positive change it can be so much more valuable to many people.
If you keep these principles in mind when creating your data visualization, I’m sure it will turn out great!
For those of you who want to dive deeper into the principles and the topic of data visualization I recommend Alberto Cairo’s books, specifically "The Truthful Art: Data, Charts, and Maps for Communication".
-Merlin
References
Alberto Cairo. 2016. The Truthful Art: Data, Charts, and Maps for Communication (1st. ed.). New Riders Publishing, USA.
