Graphical data representation

December 13, 2006 12:15 AM

Help me work out how to represent data in an innovative way.

I am an RA to an academic in the field of Education, and we’re looking at representing some data in a different way. For example, this style - what’s it called, and who invented it?

Is there a better guide to a variety of diagrams or ways of displaying data in a graphical way than Wikipedia?

Apart from Edward Tufte, who would be the leaders in the field of displaying data? Is there a name for this field? I find myself lacking in the terms to even ask this question.

I found this here:

Visual Learning
Cool software visual design

I’m also familiar with Excel and the graphs found there, and I’m wanting to move in a different direction.

posted by b33j to media & arts (12 comments total)

Not sure this is exactly what you are looking for, but you should at least checkout Swivel. More here.
posted by JPowers at 12:31 AM on December 13, 2006

There's a really interesting blog focused on data representation, called Information Aesthetics. I think you'll find it useful. And fun.
posted by Alt F4 at 6:51 AM on December 13, 2006

This has always been a subset of Technical Illustration. Any competent illustrator can give a variety ways of illustrating data in a unique way. I'd say you are best served by spending some of that grant money on a good technical illustrator.
posted by JJ86 at 7:01 AM on December 13, 2006

I think you're talking about a "tag cloud". Another regrettable term often used in this respect is folksonomy.
posted by Mr. Gunn at 9:51 AM on December 13, 2006

Also look into touchgraph.
posted by Mr. Gunn at 9:51 AM on December 13, 2006

Check out visual complexity.

The field is sometimes called information graphics (this Wikipedia entry lists other names alongside Tufte's, but I don't know whether it's a good list of leaders in the field.)

I attended a Tufte class Friday. One thing he recommended was reading the New York Times, Wall Street Journal, and Washington Post, who have all been at the representing information game a long time.
posted by Zed_Lopez at 10:29 AM on December 13, 2006

I second suggestions for looking at Information Aesthetics and Visual Complexity. Lots of examples for you to browse.

You didn't mention what kind of data you were seeking to visualize. People could suggest specific techniques or tools here if you posted more details. Or you might email the webmasters of the above 2 blogs with the details about your data. Since they have seen many types of works, they might be able to point you quickly to several good ideas.
posted by shortfuse at 1:42 PM on December 13, 2006

Thanks for all your help. Bunch of very useful answers.

Uh, there is no grant money, JJ86, and without doubt, no technical illustrators in this small town (except for this great guy who does cut outs of performance vehicles but I'm pretty sure that's not what we want).

shortfuse, it's the difference between students' perceptions of their intended career at the start of their course of study, and at the end, described with keywords. Some words are far more common than others. Some words have more positive connotations than others.
posted by b33j at 1:49 PM on December 13, 2006

Also, try contacting a certain Mr. Ben Fry. He's 1/2 of Processing, did his PhD on innovative infoviz, and made cool stuff like zipdecode which I believe was linked on MeFi at some point. Nice guy. He might point you in a suitable direction.
posted by shortfuse at 1:50 PM on December 13, 2006

b33j - Ah I just saw your reply. Forgot to preview.

So 2 sets of keywords (1 for before and 1 for after), and each keyword has a positive/negative connotation value?

First thing coming to mind is to have a tag cloud for each set.

Within each tag cloud:
- Size of tag represents frequency of occurrence. More frequent = larger.
- Color of tag represents positive/negative value. More positive = bluer; more negative = redder.

Then, lay out the tags using a combo of these 2 ways:
- Radially, with more common (i.e. larger) tags close to the center, and less common (i.e. smaller) tags away from the center.
- Symmetrically, with more positive (i.e. bluer) tags on the right side, and more negative (i.e. redder) tags on the left side.

So looking at a tag cloud, you can see (1) the relative frequency of words (position+size) and (2) the spectrum of positive/negative connotations (color).

As for before+after, you could do 2 tag clouds, then animate between them.

Off of my head, so no promises that it won't be crap. Might be a starting point.
posted by shortfuse at 2:04 PM on December 13, 2006

shortfuse, that's almost exactly what i was thinking (and kudos to you for understanding exactly what I was talking about), and I did create an inital tag cloud which was really interesting.

Here's the tricky part - each word might refer to (and I haven't yet worked out all the categories) an internal characteristic of a person in the profession (skilled), a feeling (frustration), a behaviour (aggressive), an external experience (respected) etc. So we've got with a larger point size for higher frequency, different colours for positive and negative, and then the third factor(s) of category - at that point, I think the tag's not working and I have to go somewhere else.

But this has given me much food for thought, thank you. I'll post a follow up when the paper is done.
posted by b33j at 3:37 PM on December 13, 2006

get ahold of and read this book: HiperCatalunya. excellent, excellent example of information design practically applied. Tons of information in innovative and original design applications in charts, maps and tables.
posted by ab3 at 5:16 PM on December 13, 2006