KDnuggets Home » News » 2015 » Aug » Tutorials, Overviews, How-Tos » How To Build Compelling Stories From Your Data Sets ( 15:n29 )

How To Build Compelling Stories From Your Data Sets


Are you done with digging, slicing and aggregating those numbers, your job is not over yet. Presenting your findings is an art itself, find out how by means of visualization you can achieve this.



By import.io.

Every number has a story. As a data scientist, you have the incredible job of digging in and analyzing massive sets of numbers to find what that story is. The challenge can be that while you may have an artistic bent, you may not know how to turn that beautiful visualization into something more meaningful. Is it even possible?

Even the most mundane datasets can be compelling to an audience; it’s simply a matter of presentation. This post will aim to guide you through just how you can make a statistical analysis into a compelling narrative.

Visualization Is Your Friend

From the start, visualization is already helping you to make a compelling story—so you’re starting from a winning standpoint. In fact, one study shows that people who use visual aids in presentations are 43% more persuasive in their arguments.

Now, your job is to take that visualization and make it something that’s truly compelling. To do that, we’re going to focus less on the actual visualization, and more of what’s behind it: a well-crafted story.

Create a Narrative

Whatever the dataset you’re visualizing, there’s a story that comes out of it. This can be represented in something as simple as the change over time—what is important to realize is that it’s not just numbers. The visualization isn’t simply a representation of the numbers; it’s representing a point in a larger narrative. You just need to figure out exactly what that narrative is.

Narrative Structure 101: Every Story Needs Conflict

Based on this interview from The Atlantic, it becomes clear very quickly that a compelling story hinges on conflict. There needs to be some sort of tension in the story. While that might not play out in terms of “character development” or a plot arc, there is still a way to convey this tension—that something is wrong, or broken, or being fixed. There is significance to the data beyond it simply presenting something new.

The Different Types of Plots

According to Christopher Booker, there are seven basic plot types: overcoming the monster, rags to riches, the quest, voyage and return, comedy, tragedy, and rebirth. Most commonly, we see overcoming the monster—but we don’t get the full story. That’s the beauty of data visualization: you don’t have to tell the story, but you have to present some sort of tension that compels your audience to dive into your visualization.

Screen Shot 2015-08-20 at 9.12.06 AM

In this video, surrounding U.S. gun death statistics, the monster is clearly gun violence. They do not present a solution, but rather simply show us the monster. But it’s not just the monster that makes this video compelling, they include several other narrative elements that draw the audience in.

Identifying The Narrative Elements

The five main elements of a narrative are the character, setting, conflict, plot, and theme. In the above example, it’s extremely easy to identify every single one: the characters are the victims of gun violence; the setting is the U.S.; the conflict is that they’re losing years they could have had; the plot is that every day, someone in the U.S. is losing their life to gun violence; and the theme is that gun violence in the U.S. is stealing lives.

They do not present a solution, that’s for the audience to conclude themselves, but rather than simply presenting the statistic that 9,595 people were killed because of gun violence, totally 413,342 stolen years, they used a beautiful visual presentation of each life up to the point of death and then the years that were stolen to make the numbers both tangible and significant.

Build On Your Story

The challenge for most data storytellers, however is that they’re not working with “compelling” data. You could be working with cell phone customer data in China, or consumer behavior based on e-commerce search queries. So how do you make that into something persuasive and beautiful?