Three basic things about data visualization

Over the course of 2019, I gave several talks about data visualization. The content morphed over time as I searched for one core message to leave folks with. Recently, it struck me that I’d not been focused on one thing, I’d been focused on three. Granted, there are many reasons to think about data visualization and folks could easily add to these three items. But, for me, these basic concepts make a quick, powerful argument for the value proposition of visualization.

Visual space

Take a moment and think about the physical dimensions of whatever device you’re reading this on. Your brain will throw up a minor speed bump because it’s doing a bit of work that you might not notice. It takes a moment, but our eyes will expand whatever has our attention in such a way that it fills up most of our visual perception. This rectangle is the space that’s available to you to convey a message and the amount of space doesn’t change! Yes, we can zoom in or out, and we can scroll, but this imposes a burden on the viewer. They’re navigating a new geography over which you have no control. And the fact of the matter is, that no amount of scrolling or zooming will magically change a single sheet of paper into giant canvas. You have to work with what you’ve got.

Now. Let’s assume that you’re filling that rectangle with a table. Font size means that we are instantly constrained with the amount of data that we can present.

The essential truth is this: words, tables and plots operate according to the same rules of visual space. But plots scale in ways that other presentation simply can’t.

Order of magnitude

The base ten numbering system means that we need additional space whenever we crest an order of magnitude. In other words, once we move from 9 to 10, we need to represent two digits. The number 10 takes up (roughly) twice as much space as the number 9. The consumption of visual space with respect to order of magnitude is arithmetic, rather than geometric.

I think this has cognitive problems. Odds of 10 to 1, take up as much visual space as odds of 99 to 1. But those are very different odds! Now, we’re not

Geometry beats semiotics for comparisons

Speaking of Arabic base 10, it’s not great for comparisons. Which is larger 9 or 11? How about 9 or 1.1? OK, this isn’t hard, but consider how much work your brain has to go through. It’s not enough to simply count the digits. If we did, we’d think that 1.1 is larger than 9. We have to map our concept of quantity and counting to a symbolic representation.

I’ll aim to have further posts about each topic individually in the next few weeks. Look forward to sharing those and — as always — I love to hear comments or questions!

Brian Fannin
Brian Fannin
Captain

I’ve been an actuary for over 20 years an Associate of the CAS since 2002 and a Certified Specialist in Predictive Analytics (CSPA) since 2017. I joined the staff of the CAS in March of 2018 as Research Actuary, where my focus is to enable CAS committees and research partners to work efficiently in developing relevant, practical content.