[MUSIC] Uncovering information with data, numbers, and metrics is not always sufficient to convey your results to others. For people to take action on insights gained from data science, they must first understand the implications of those findings. Communicating results to a non technical audience, often referred to as data storytelling, is an important skill. Good data storytelling combines solid data science, clear visualizations, and a meaningful narrative. You've practiced the data science component in depth. So let's now focus on the other two components, starting with visualization. To illustrate this concept, let's use the storm events dataset from exploratory data analysis. In the Course One project, you analyze data from Hurricane Harvey to identify where to send claims adjusters. Now imagine you're communicating analysis to recommend increasing the cost of insurance due to the hurricane. Showing the cost of Hurricane Harvey compared to other events is a good way to illustrate the magnitude of the damage. While something like this plot does show that information, it also includes the location of events. This extra detail obscures key information and can overwhelm your audience. Effective visualizations display only what's necessary for the audience to understand your main points. In this case, a bar or pie chart is more useful for comparing the damage caused by Hurricane Harvey to all other events. Good visualizations also include a title and labels so your audience can quickly recognize the information being presented. The goal is to create figures your audience can grasp without much effort. You can accomplish this by using pre attentive attributes. Pre attentive attributes include size, color, shape, and other properties that add contrast and trigger unconscious reactions. You've actually been using pre attentive attributes throughout this specialization. As you start creating your own reports, make sure to create your visuals with properties that introduce sufficient contrast. The third element of data storytelling is crafting a compelling narrative. There's no single recipe, but there are several aspects to include. Your narrative should clearly state the problem, the current situation, a proposed solution, or a recommendation, a conclusion, and next steps. How you present each topic will depend on your audience. Using the Hurricane Harvey example, does the business team already know that insurance rates need to increase? If not, you'll need to describe the current situation and convince your audience of the problem. You also need to tailor your presentation to the technical background of your audience. For example, when recommending the new rates, a technical audience may want to see a histogram of likely outcomes. While you may present the results more simply to a general audience. As you construct your narrative, look for opportunities to make the story concise, compelling, concrete, and credible. Keep the story concise by focusing on what's most important for the audience to know. Avoid extra details that can distract the audience from your main point. A compelling narrative will keep your audience's attention. Introducing surprise or emotion are two ways to do this. For example, highlighting that in two weeks, Hurricane Harvey caused more than twice the damage of all other events combined, may surprise the business team and get their attention. Make your narrative concrete by using specific examples. As a data scientist, your recommendations are based on the full data set. However, highlighting specific observations make the story more relatable to your audience. Finally, make your story credible. Draw conclusions based on good data science. Use established tools and resources to support your claims. And discuss limitations or unknowns in your model. Such as features you lack or missing data. In summary, data storytelling is the ability to convey your results as a story your audience can easily follow. There are many ways to tell a good story, but they all have clear and purposeful visualizations. A well thought out narrative. And of course, good data science. With this ability, you can keep your audience engaged, informed, and inspire them to take action. [SOUND]
