I recently taught a course on #dataviz. As someone who was interested in the subject but had not formally learned it before, I learned a lot along the way! I'm going to be sharing some fun and useful tidbits I learned here.

not gonna lie it was a little disturbing to learn that a large amount of the post-apocalyptic jargon in Mad Max / Furiosa is actually just regular Australian slang

I've had a wonderful time hanging out with my fellow @UNSWScience STEMM champions and @AstroKirsten today!

I am begging Australians to learn how to say "El Niño" and "La Niña" correctly.

Celebrate #IWD2024 with us! We introduce you to 25 rising stars in #DataScience: 5 hours/5 women each hour 📌Online 📆Tues, 12 March 🕛12pm-5pm (MEL, SYD, CBR) 🕦11:30am-4:30 pm (ADL) 🕚11am-4pm (BNE) 🕘9am-2pm (PER) 🕘9pm, 11 Mar: USA EDT https://t.co/XWCJIWLrme #womeninSTEM

Posts 13-15 are from "Data Visualization: A Practical Introduction", by Kieran Healy. https://t.co/GdpcIQGDPE 19/19

Personally, I appreciate having learned about Tufte's principles so that I look at plots with a more critical eye. I try to consider whether each element in a plot is necessary or not--even if I do end up keeping some unnecessary elements in the end. 18/19

(For different gridline styles, try different ggplot themes: theme_classic(), theme_minimal() and theme_tufte() remove gridlines entirely, while many newspaper themes remove some gridlines: theme_wsj(), theme_economist(), bbc_style() are a few examples) 17/19

So, how much should you simplify a plot? There's no easy answer. For example, take gridlines. Looking at recent visualizations from the New York Times, some plots have no gridlines, some only have a few important gridlines, and some have a full set of gridlines. 16/19

Even though it is a duck, it gets across the message in an effective and memorable way. Bateman et al., 2010 found that more embellished infographics are more memorable than plain statistical graphs. It seems there are tradeoffs for emphasizing simplicity. 15/19

Or consider the famous graph "Monstrous Costs" by Nigel Holmes. It certainly contains chartjunk--in fact I would go as far as to say it is a duck! 14/19

Which do you prefer? Anderson et al. (2011) found that Tufte's boxplot is more cognitively difficult to process than a traditional boxplot. 13/19

In fact, there are some extraneous elements there. Here we have Tufte's version of a boxplot instead. 12/19

Consider a classic boxplot. Is the data-ink ratio maximized? 11/19

A related concept introduced by Tufte is the data-ink ratio: the proportion of a graphic’s ink devoted to the non-redundant display of information. 10/19

Here's a "duck", which contains confusing 3D and meaningless colors. He points out the graph only represents 5 points of data, given the percentages must add up to 100. He comments that "this may well be the worst graphic ever to find its way into print." 9/19

"...then that graphic may be called a duck in honor of the duck-form store, 'Big Duck'". The Big Duck is a building in New York that was built by a farmer to sell ducks and duck eggs (wikipedia). 8/19

Finally, what is "the duck"? "When a graphic is taken over by decorative forms or computer debris, when the data measures and structures become Design Elements, when the overall design purveys Graphical Style rather than quantitative information..." 7/19

As to background gridlines, Tufte held a strong opinion: "one of the most sedate graphical elements, the grid should usually be muted or completely suppressed so that its presence is only implicit." 6/19

Here's an example of an unnecessary texture/pattern. 5/19

Chartjunk abounded in early days of Excel-generated graphics. Think of unnecessary 3D representations and loud patterns. Tufte highlighted three common types of chartjunk: texture/pattern, grids, and "the duck". 4/19