May 2016

In This Issue

Quick Links

The Who of Data Visualization

campbell-mary.jpg By Mary Pat Campbell 

In my last article on data visualization, I set up a series of articles covering this subject: 

  • The Why of data visualization—questions to ask when visualizing numerical information,
  • The Who of data visualization—major figures and books in advocating data visualization best practices,
  • The Where of data visualization—websites to polish your data visualization game,
  • The What of data visualization—software to implement data visualization, and
  • The How of data visualization—specific data visualization techniques to consider in actuarial practice.

(The When of data visualization being NOW, of course.) 

For this article, I’m going to concentrate on major figures in data visualization—people who have multiple books, courses, and other materials in which you can learn to improve your data visualization approaches.

Edward Tufte

Tufte is the modern prophet of good visualization techniques, though he does not provide much practical information on how one goes about making the graphs. That’s fine, because people later in this article do give you the tools to make certain kinds of specialized graphs and displays.

My main recommendation with respect to Tufte is: buy his books (listed below, with links to their Amazon pages).

Tufte’s Resources

Books

Links

I love Tufte

Tufte’s books are gorgeous. (Buy the hardcover versions—the paper is luscious … if they’re using the same paper as more than a decade ago.) However, I use them more as inspiration than as reference texts. 

Yes, Tufte gets into the nitty-gritty of design principles, but more than anything else, he has example upon example of gorgeous data visualizations. (Yes, also some stinkers (on purpose)) His prose style is lacking, but I find the specific examples far more useful than words describing them. 

I own all of his books (yes, even the short one bashing PowerPoint). Thanks to Amazon’s record-keeping, I realize I bought my first Tufte book back in 2003, as I was embarking on my actuarial career. I have gone back to it often to think things through. 

The first Tufte book I bought is the primary one I recommend for actuaries: The Visual Display of Quantitative Information. The reason this is the one I recommend is that many of the other books are about conveying non-exactly-quantitative information visually, much of which has little to do with the sort of thing we do as actuaries. While I enjoyed the inset example in Visual Explanations from Humphry Repton architect’s plan to alter a cottage, taking me back to toddler-hood and pop-up books, this sort of communication is not used so often when communicating about mortality trends, for example. 

However, Visual Explanations has two examples that may be of interest to actuaries. First, the famous map created by John Snow in 1854, showing the incidences of cholera in the Broad Street area of central London. Taking a list of 83 deaths, and a local map, he placed a bar for each death and its location. The deaths showed the highest number near the pump that turned out to be the source of contamination; investigating deaths that occurred farther from the pump showed that all the people stricken had used the pump regularly. With this map and his additional findings, one week after cholera broke out, a pump handle was removed and the epidemic stopped. This is an example of an effective visual display, in both discovering the source of a problem and communicating that information to other people. 

The second example of interest is the Space Shuttle Challenger explosion of 1986. Tufte gives examples of the communications actually made by engineers who were trying to warn about the danger of O-ring erosion in low temperatures, and critiques these communications. In some cases, numbers were simply given as tables, hindering any kind of pattern discovery. Other reports showed erosion results, but in a confusing way that was hardly better than the tables. Tufte shows a couple different presentations he thinks would have been more convincing to the people who had to decide whether to launch that cold January morning. 

A final note: do not be put off by his website, http://www.edwardtufte.com/. He is definitely more oriented toward book-type and report-type layouts. 

However, for one willing to look past bad web design, there is good content. In particular, there are some discussion threads, often jumping off from his books’ content. Also, some pretty posters. 

Tufte also conducts one-day courses, which comes along with four of his books. Tufte is based out of Connecticut, so most of these courses are on the East Coast. I’ve never been, but I had one friend who did—he said that you pretty much got the same benefit from just reading the books.

Nathan Yau, FlowingData

Different from Edward Tufte, Nathan Yau has a very nicely designed website, with a membership feature. I do not have a membership, so I cannot comment on that. I have often visited Yau’s site and checked out his blog posts; also I own one of his books. His books are fine, but I recommend his web material more than the books. 

Also unlike Tufte, Yau gets a little more into how to create the graphs he shows. However, these are not step-to-step guides. My understanding is that he has more involved tutorials in his members-only part of the website. Most of Yau’s statistical work is done in R, and sometimes he uses R for the graphics, and sometimes JavaScript for interactive graphics. He has said before that most of his tutorials involve using R. 

Yau is far more modern than Tufte (who is more given to showing really old historical examples—cool stuff, but I don’t need to tell the next Napoleon not to invade Russia in winter.) Yau gives me ideas on some ways to present information, though this is not so much reference for me.

Yau’s Resources

Books

Links

Stephen Few, Perceptual Edge

Stephen Few and his consultancy Perceptual Edge may provide the most comfortable resources for actuaries, as he is one of the few people in the “serious” dataviz community who will give examples in Excel. 

Indeed, it looks like most of Few’s examples were made in Excel (albeit in nice color schemes, and other alterations from default settings to make the graphs look better). 

I have two recommendations:

  1. Subscribe to his Visual Business Intelligence Newsletter
  2. Get his text Show Me the Numbers: Designing Tables and Graphs to Enlighten (2ndedition)

The first is a no-brainer. It’s free. It comes quarterly, and most of his examples relate to a business, as opposed to an academic, environment. Also, pretty much all of his finished examples can be made in Excel, so one isn’t worried about having to learn R or some other new tool. 

The text is probably a bit basic for more seasoned actuaries, but is a good text for orienting entry-level people. That said, the text can also help more senior people develop exemplars for information display. Having a set style for communicating quantitative information can be useful. And I do love the chapter “Silly graphs that are best forsaken” near the end of the book.

Few’s Resources

Books

Links

Others?

As I got to my own exhaustion, I left off Alberto Cairo (check out his book The Truthful Art: Data, Charts, and Maps for Communication ) and David McCandless (check out his site Information is Beautiful http://www.informationisbeautiful.net/) … but even throwing them in, I know I’m leaving off others who can also help in improving our data visualization skills. 

Are there any people or resources you recommend for dataviz that I didn’t cover? Please email me at marypat.campbell@gmail.com … or write an article for CompAct! 

Mary Pat Campbell, FSA, MAAA, is a vice president, Insurance Research at Conning in Hartford, Conn. She can be reached at marypat.campbell@gmail.com.