Join Doug Rose for an in-depth discussion in this video Metaphors, part of Learning Data Science: Tell Stories With Data.
- We live in a world filled with metaphors. You'll see them in common phrases. You can be as busy as a bee, heartbroken, or as quiet as a mouse. They're common in literature. You've probably heard that all the world's a stage. Or that Macbeth tried to see the seeds of time. Politicians use them in their speeches. There was President Reagan's morning in America, and President Obama saying the economy had run into a ditch. The reason you use metaphors is that they work. They connect something you know to something you don't know.
They make the unfamiliar seem more familiar. In their book, Metaphors We Live By, by George Lakoff and Mark Johnson, they argue that metaphors are essential to how we think. We use metaphors to understand concepts, such as love, war, and cooperation. They said that people who impose their metaphors on culture, get to define what we consider to be true. When you're telling a data science story, you should use metaphors as a way to ease in new ideas.
They make the unfamiliar seem familiar. When you hear a story about something familiar, you're more likely to connect it to some meaning. From a literary perspective, a metaphor is making two things the same. Like the phrase, chain reaction. You'll think about how a chain works with each link tied together, then you think about something happening. The one thing reacts with several other things. This is much easier to imagine than a term like self-amplifying event.
In storytelling, just think of metaphors as anything that connects an unknown to something you know. That way you don't have to worry too much about the nuances between metaphors, allegories, similes, and analogies. Just keep it simple. If you're equating two different things, then think of it as a metaphor. Data science has a lot of difficult concepts, so you already have a few well-established metaphors. There's data warehouses, data mining, and data lakes. Remember, our early metaphor about panning for gold.
This is the type of poetic language you'll want to use when describing difficult data science concepts. Imagine your team is working for one of the big movie studios. You're trying to figure out a way to use predictive analytics to decide how many screens you want to show a new movie. You don't want to show a movie on too many screens, then you'll have a lot of empty seats. You also don't want to show it on too few screens, then the people won't be able to get tickets, or skip the show entirely. Your data science team has gathered up structured and unstructured data.
They have a lot of structured data that showed people clicked on the movie trailer, on many different websites. You also have a lot of unstructured data, that shows that there's a lot of movie buzz. People are talking about your movie a lot on social media sites, such at Twitter and Facebook. When you tell your story, you don't want to use terms such as unstructured data analysis. Instead, you want to use metaphors, such as coffee shop chatter, or movie buzz. That way the audience immediately knows the value of the data.
They'll know the source, and they'll have an idea of how it was created. Think about how these two stories sound. Imagine if you said our unstructured data analysis suggests that there's a lot of interest in this title. Now imagine saying, a lot of the entertainment buzz online suggests that people really want to see this movie. You could use metaphors in other ways, as well. You might want to say these are hot tickets, or after a few weeks, there might be a cool off period. All these metaphors make the story more interesting and fun.
That will keep your audience engaged, and help them find some meaning. When you use metaphors, you're likely to break down the barrier between you and your audience. In data science, there's some danger in using complex terms. You run the risk of creating a wall between you and your audience. A metaphor not only makes the story sound more interesting, it also lowers the barrier to participate. Your audience might be more likely to question the value of movie buzz. These same people might not ask questions about unstructured data analysis.
The more your story engages the audience, the more likely they are to extract some meaning.
- Structuring a data science story
- Defining plot, conflict, and details
- Going beyond reporting
- Knowing your audience
- Working with data
- Introducing visuals
- Eliminating distractions
- Incorporating metaphors
- Motivating the audience
- Avoiding pitfalls