With the new “one reading and discussion per week, but it’s spread out over the week” style, the numbering system isn’t as meaningful. But, there are some readings for the week: some to do for Monday (before Tuesdays class), and others you can do later in the week (before Thursday’s class). You need to make an initial posting about the first readings for Monday, and more discussion about the other readings.
Due Dates: Initial readings and postings, Monday, March 16, 11:59pm (before class Tuesday, March 17).
Turn-in Link: Reading 16 & 17 on Canvas
The goal of this reading is to give you some ideas as to what people try to do to deal with high dimensional data. Part of the answer is “a whole bunch of not-so-great things.”
The requirements are the Ware paper and one of the two older papers. And to look around on the web for a parallel coordinates example and a scatterplot matrix example. (all described below)
Update (3/12): I wasn’t clear what was for Tuesday, and what was for Wednesday. Given the other things going on in class, you probably want to do most of the reading (at least the required stuff) for the initial posting. At least do the Ware reading and the “old high-dimensional paper” reading. Do the parallel coordinates and scatterplot matrix readings before Thursday’s class.
For discussion: in your initial post (and discussion) describe the challenges of showing a bi-variate field, and comment on how the solutions you’ve seen (Ware describes a few) address it (or not). After reading looking over the older high-dimensional paper, comment of how the problems addressed are different. In this discussion, please discuss the various ways of showing high-dimensional data, and relate them to some of the things we’ve already discussed in class. (hint: many of the things we’ve discussed are high-dimensional – like using multi-channel glyphs).
One specific problem is dealing with 2 variables (bi-variate) over a 2D fields (like temperature and wind velocity at every point on a map). For this part, you need to read:
- Colin Ware, “Quantitative Texton Sequences for Legible Bivariate Maps,” IEEE Transactions on Visualization and Computer Graphics, vol. 15, no. 6, pp. 1523-1530, Nov./Dec. 2009, doi:10.1109/TVCG.2009.175 (ieee page, colin’s version)
This paper is really enough, since it gives good reasons why all of the previous techniques are unsatisfying. That said, students in previous editions of the class got a lot out of these other papers – in particular, 3 inspired a bunch of things. 4 is actually a much broader paper that is useful for many different things. All of these are optional.
- Bruce E. Trumbo. A Theory for Coloring Bivariate Statistical Maps. The American Statistician, Vol. 35, No. 4 (Nov., 1981), pp. 220-226 (web version)
- James R. Miller, “Attribute Blocks: Visualizing Multiple Continuously Defined Attributes,” IEEE Computer Graphics and Applications, vol. 27, no. 3, pp. 57-69, May/June 2007, doi:10.1109/MCG.2007.54 (ieee page)
- Haleh Hagh-Shenas, Sunghee Kim, Victoria Interrante, Christopher Healey, “Weaving Versus Blending: a quantitative assessment of the information carrying capacities of two alternative methods for conveying multivariate data with color.,” IEEE Transactions on Visualization and Computer Graphics, pp. 1270-1277, November/December, 2007. (ieee page) (healy’s version)
- Daniel A. Keim, “Designing Pixel-Oriented Visualization Techniques: Theory and Applications,” IEEE Transactions on Visualization and Computer Graphics, vol. 6, no. 1, pp. 59-78, Jan.-Mar. 2000, doi:10.1109/2945.841121 (ieee page) (author’s page)
For multi-variate techniques, there were a bunch of terrible old ideas that didn’t stand the test of time. The only (general purpose) ones that did are parallel coordinates and scatterplot matrices. So, I would like you to “read” an old paper to get a sense of the zoo of ideas that went away, and then to get a bit more of an appreciation for those that survived.
For the old papers, you can pick one (or both) of:
- Wong and Begerton. 30 years of Multidimensional Multivariate Visualization (find it here). 1997
- “High-Dimensional Visualizations,” G. Grinstein, M. Trutschl, U. Cvek, 7th Data Mining Conference-KDD 2001, San Francisco, California, 2001. [PDF]
The former paper is more archaic (and has some really crazy designs – hard to believe that people actually used them). The latter of the two hints at some of the analytical methods for dealing with high-dimensional spaces.
To learn about scatter-plot matrices and parallel co-ordinates, there really aren’t readings. Instead, what I’d like you to do is to see how modern systems deal with some of their short-comings through clever design and interaction. If you really want, you can read the original parallel coordinates paper (here), but it’s not so helpful.
For scatterplot matrices, check out the Scatter-dice system (project page). Rather than reading the paper, you can watch the video. The video that comes up afterwards on YouTube (The FlowVizMenu and Parallel Scatterplot Matrix) also shows off these ideas – in fact it mixes SPoMs and Parallel coordinates.
For parallel coordinates, I have no great suggestions. There is a nice demo of a D3 implementation (here) that has a lot of nice features (play with the demo and see how selection and coloring makes parallel coordinates more palatable). Be sure to try using 2 filters.
If you find any other great examples on the web, please point them out in the discussion.
If you want to see a generalization of parallel coordinates, you can read
- Claessen JHT, van Wijk JJ. Flexible Linked Axes for multivariate data visualization. IEEE transactions on visualization and computer graphics. 2011;17(12):2310-6. Available at: http://www.ncbi.nlm.nih.gov/pubmed/22034351. Accessed March 9, 2012. (this reading is optional for class)
I think this paper might be here to prove that not all of Jarke’s ideas are great.
