[Sigia-l] Information Visualization

[Eric Reiss]

I tuned in late. Pardon me if this has all be said before by wiser
participants.

It strikes me that visualization is always (and only) the means to an
end – understanding the information. It’s not the end itself. That’s
why some fancy electronic presentations are less illuminating than a
few well-considered finger sketches in the sand. As far as I’m
concerned, the technology is entirely secondary as long as people
“get the message.”

Like Karl, I’ve been looking through the literature. Frankly, I’m
surprised the word “understanding” doesn’t turn up more prominently
in definitions that otherwise promote concepts such as “discoveries,
decisions, explanations.” (see
http://www.cs.umd.edu/hcil/pubs/presentations/FutureWeb/sld020.htm
for one of my least-favorite definitions.

For me (and I suspect many others), most visualization falls into
five main categories:
Displaying
Comparing
Explaning
Combining (two variants)

1 - To display a lot of homogeneous info (into
charts/graphs/topological maps/whatever)

These types of visualization help us to plot historical data, gain
perspective, and spot trends in the best George Santayana tradition.
(“Those who do not remember history
”) Page 13 of Tufte’s “Visual
Display of Quantitative Information” has some good ground rules.

2 - To compare two or more types of information (age of employee vs.
days sick)

This visualization is designed primarily to highlight interdependency
and/or coincidence. (A Danish artist/philosopher, R. Broby-Johansen,
once compared hemline height and the U.S. GNP from 1913-1953 with
surprising results).

3 - To explain a process – often linear (like how to put your Ikea
furniture together or the biochemistry of 5-ASA in the treatment of
Crohns Disease.)

Here, you’re likely to see lots of illustrations showing Tab A being
inserted into Slot B, for example. Tufte has pretty pictures in
“Visual Explanations.”

4 - To combine several interdependent pieces of information in which
there are no variables (combining weight and height to calculate BMI)

Not strictly visualization since the numbers are often only
interesting as an academic exercise. The result is the main point
(I’m too fat. I’m OK. I don’t care about the math.). Of course, if
you create a chart to show the basic relationship, you’ve really gone
back to number 1.
 
5 - To combine several interdependent pieces of information in which
there are one or more variables. (charting contributions to a private
pension fund – monthly contribution is a variable, as is age of
retirement, percentage of salary, etc.)

In this instance, the idea of visualization is often to show how
individual variables contribute to/affect the results, thus making
the process more understandable. (A Danish pension fund did exactly
this to great advantage. Check out:
http://www.pfa.dk/sw1017.asp?noget=andet&master=0. From left to
right, the sliders read: “Monthly salary,” “Current age,” “Retirement
age,” “Savings,” “Percentage of income to be put into retirement
fund.” At the upper left, Kvinde = woman, Mand = man. At the upper
right, you can choose the payout period – from 10 years to rest of
life. The bar in the middle explains how much you’ll receive in
retirement benefits each month. The gauge at the top shows you the
percentage of your current income that will be paid out each month
when you retire. The higher the better.)

Karl is looking for a framework to guide development. Great. But
until people can agree on the goal of visualization, this probably
ain’t gonna happen. Amazing that so many think it’s all about
information retrieval. Or maybe I just got it all wrong


Cheers,
Eric
- - - - - - - - - - - - - - -
Eric Reiss
e-reiss aps
copenhagen, denmark