On the more formal side of things

When presenting data findings to a more formal audience, I like to stick to the following six steps:

1. Outline the state of the problem: In this step, we go over the current state of the problem, including what the problem is and how the problem came to the attention of the team of data scientists.
2. Define the nature of the data: Here, we go into more depth about who this problem affects, how the solution would change the situation, and previous work done on the problem, if any.
3. Divulge an initial hypothesis: Here, we state what we believe to be the solution before doing any work. This might seem like a more novice approach to presentations; however, this can be a good time to outline not just your initial hypothesis but, perhaps, the hypothesis of the entire company. For example, “We took a poll and 61% of the company believes there is no correlation between hours of TV watched and work performance.”
4. Describe the solution and, possibly, the tools that led to the solution: Get into how you solved the problem, any statistical tests used, and any assumptions that were made during the course of the problem.
5. Share the impact that your solution will have on the problem: Talk about whether your solution was different from the initial hypothesis. What will this mean for the future? How can we take action on this solution to improve ourselves and our company?
6. Future steps: Share what future steps can be taken to address the problem, such as how to implement a solution and what further work this research has sparked.

By following these steps, we can hit on all of the major areas of the data science method. The first thing you want to hit on during a formal presentation is action. You want your words and solutions to be actionable. There must be a clear path to take upon the completion of the project, and future steps should be defined.

The why/how/what strategy for presenting

When speaking on a less formal level, the why/how/what strategy is a quick and easy way to create a presentation worthy of praise. It is quite simple, as follows.

This model is borrowed from famous advertisements – the kind where they would not even tell you what the product was until there were 3 seconds left. They want to catch your attention and then, finally, reveal what it was that was so exciting. Consider the following example:

Hello everyone. I am here to tell you about why we seem to have a hard time focusing on our jobs when the Olympics are being aired. After mining survey results and merging this data with company-standard work performance data, I was able to find a correlation between the number of hours of TV watched per day and average work performance. Knowing this, we can all be a bit more aware of our TV-watching habits and make sure we don’t let it affect our work. Thank you.

This chapter was actually formatted in this way! We started with why we should care about data communication, then we talked about how to accomplish it (through correlation, visuals, and so on), and finally, I am telling you the what, which is the why/how/what strategy (insert mind-blowing sound effect here).

Summary

Data communication is not an easy task. It is one thing to understand the mathematics of how data science works, but it is a completely different thing to try to convince a room of data scientists and non-data scientists alike of your results and their value to them. In this chapter, we went over basic chart making, how to identify faulty causation, and how to hone our oral presentation skills.

Our next few chapters will really begin to hit at one of the biggest talking points of data science. In the last nine chapters, we spoke about everything related to how to obtain data, clean data, and visualize data in order to gain a better understanding of the environment that the data represents.

We then turned to look at basic and advanced probability/statistics laws in order to use quantifiable theorems and tests on our data to get actionable results and answers.

In subsequent chapters, we will take a look into machine learning (ML) and the situations in which ML performs well and doesn’t perform well. As we take a journey into this material, I urge you, the reader, to keep an open mind and truly understand not just how ML works, but also why we need to use it.