Where Cole in The Sixth Sense sees dead people, Isaac Reyes sees bad charts. In his role as founder and data scientist at DataSeer Inc., Reyes helps individuals and businesses better understand and convey the messages hidden in data. And in his keynote at ODSC East 2018, Reyes shared the key principles of data storytelling – the art of turning slide decks chock full of numbers and charts into data stories that are clear and impactful. Here were some of his most trenchant insights:
Having placed the audience at the center of your data story, it’s time to start creating and refining the visualizations which will carry the narrative. Reyes’ approach draws on the Gestalt laws of visual perception, named after the famed German Gestalt psychologists like Kurt Koffka and Max Wertheimer, who sought to understand how humans acquired and maintained meaningful perceptions through sight. Two such principles that are particularly relevant to data visualizations include:
- The Law of Proximity, which says that things that are close together seem to go together,
- The Gestalt Law of Similarity, which holds that things that are colored the same appear to go together
Thus, color and spacing ought to be used strategically to help deliver the message of the data story at hand. In particular, careful use of colors and weightings can distinguish the most important chart elements from the rest. A problem with common graphing tools is that they give equal color importance to every data unit. But think of your favorite literary tale; does every character have the same importance? Finally, Reyes channels Edward Tufte, the statistician who pioneered many modern techniques for data visualization, in urging clients to remove unnecessary elements or “chart junk.” Any ink that’s there that’s not showing or contextualizing the data, Reyes suggests, is very likely superfluous.
Bringing it all together, Reyes shows how a messy plot attempting to convey a message about Apple’s revenue growth in China can be transformed into a clear, concise, and impactful data story:
Slide copyright Isaac Reyes, ODSC East 2018
With a knowledge of these principles and some practice, Reyes suggests that all data scientists can learn how to tell great data stories.
-Data Storytelling is the process of using data, visualizations, and narrative tools to send messages and communicate insights about data.
-Effective data storytelling revolves around one’s target audience, best thought of as the protagonist of the story being told. For example, the best titles for charts and visualizations depend on the audience and what they need to know.
-Design principles for coloration and spacing based on the Gestalt laws of perception help guide an audience through a data story.
-Always be wary of “chart junk,” or elements that don’t add to the message behind a chart.