Data Science: The Hard Parts

Chapter 7. Narratives

You’ve spent weeks working on your project and are now ready to present the results. It feels like you’re almost done, and just have to deliver the output.

Many data scientists think this way, and put little to no effort into building compelling narratives. As described in Chapter 1, to have end-to-end ownership, it is critical to persuade your stakeholders to take action with your results. This type of extreme ownership is critical to create value; hence, you must master the art of storytelling.

There are plenty of resources out there to learn about storytelling (I’ll suggest some at the end of this chapter). This chapter builds on that body of knowledge, but I will deviate slightly to highlight some skills that are specific to data science.

What’s in a Narrative: Telling a Story with Your Data

Using a standard dictionary definition, a narrative is just a sequence of connected events. These connections make a story. I will enrich this definition by saying that it should also accomplish an objective.

What is the objective that you want to achieve? In general narratives, it could be to persuade or engage. These apply also to data science (DS), of course, but most importantly, you want to create value, and for that you need to drive actions. A successful story should help you accomplish this objective.

Let’s reverse engineer the problem and identify conditions that help us achieve this:

• Clear and to the point
• Credible
• Memorable
• Actionable

Clear and to the ...