If you ask a room full of people with varying backgrounds in data science, graphic design, analytics, or any other assortment of skills and experiences for the definition of “data visualization,” chances are you are going to receive just as varied a response as the personalities in the room. Chances are, too, that you will also uncover some resonant themes and understandings, as well as some very contrary opinions. If you do a quick Google search online, or ask a number of leading minds in the data visualization space, you will likely get a similar result. As with any type of tool or principle, a level of disagreement is to be expected—encouraged even, as a consequence of greater discussion is (generally) more engagement, clarity, and thoughtful consideration into supporting components. Even in my journey, I have worked with companies and data scientists who are building visualizations so scientific and its almost hard to believe that they are data visualization. At the other end of the spectrum, I have also seen organizations where graphic designers are responsible for building data visualizations that sometimes are more suitable for artistic display, than in any corporate boardroom. I have seen dashboards both glorious and banal. I have even met people who have metaphorically lived by data visualization, and others than snub their nose at it. When it comes to any area of data, we are a mightily opinionated group.

Perhaps the most quintessential of examples of the power of data visualization comes from what is known as Anscombe’s Quartet, four seemingly identical data sets constructed in 1973 by statistician Francis Anscombe. The data sets, shown in a table (Figure 9.1), appear to be identical when compared by their summary statistics. Each set has the same mean of both X and Y, the same sample variance, the same correlations between X and Y in each case, the same linear regression, and so on.

While there is a vast host of different visualization types, there are generally three categories of visualizations. Each serves a different purpose, and because of this, it is important to understand the distinctions between each category and how they can be used to accomplish the goal for the data visualization. We will explore each type (with a relevant example for context) in the subsequent sections (Figure 9.5).

One of the best depictions of hybrid data visualization can be found in the infographic.

In chapter: Improved Agility and Insights Through (Visual) Discovery, a conversation on the forms of visual discovery, we reviewed—at a very basic level—traditional forms of visual discovery (forms that provide simple, straightforward visual representations of data) and the more innovative visualizations (like heat maps or network visualizations) that take the next step to advancing visual data discovery through visualization.

It was data visualization pioneer Edward Tufte (1983)—referred to by The New York Times as “the da Vinci of Data” (Shapley, 1998) who wrote, “Graphics reveal data. Indeed, graphics can be more precise and revealing than conventional statistical computations.”

This chapter concludes the depth of discussion that we will take on the mechanics and best practices (or, golden rules) of data visualization. More resources can be found in the Appendix. The point here was simply to provide a distilled look at some of the criteria for successfully crafting an accurate and appropriate visualization, and beginning to understand the plethora of options available and the undercurrent of meanings and uses behind them, as well as how some of the visual design principles of the previous chapter are applied and affected when putting data at the forefront of a visual conversation.