The problem of displaying a data set that is multi-variate, time-varying, and comparative in nature is inherently very difficult. Jonathan Woodring and Han-Wei Shen at Ohio State have developed a very aesthetically pleasing method of displaying such information using three dimensional color mapping. However, their solution leaves a lot to be desired in terms of the composition of a successful visualization. While the colors do make for very visually pleasing images, the visualizations themselves are very difficult to interpret. The value of any given data point is encoded in its color and its position is representative of its relation to other data collections as defined by a set vocabulary of logical operations that the tool can visualize (over, in, out, atop, and xor). In addition to the limitation on the amount of data that is provided, the visualization provides the user with neither manner of inferring what the values encoded by the color actually represent nor a physical representation of the meaning of the positioning coordinates. Essentially, the user has no manner of extracting any data out of the visualization other than two points being different over one dimension or another. A user simply looking at the output of this visualization cannot really gain much from the end result other than a pretty picture generated from a complex data set. This visualization technique seems to have potential, but in it’s current state, it is simply not a very useful mechanism for data comparison.
Above is an image of a visualization generated by this tool representing a logical combination of data points from the Supernova Initiative Data Set. The paper and additional images can be found here.
{ 2 comments }
I am in complete agreement with Danielle in that the above visualization is a classic example of “cool, but what for?” There is something to be said about a visualization that can be complete in and of itself, and doesn’t require additional explanation, worse, a complete conference paper for explication. The posted image has the following caption in the paper, “Tera-scale Supernova Initiative data set visualizing Catop(A in B).” It is entirely likely that if I were to read the paper I would know what the above visualization is about, but I shouldn’t have to. The purpose of visualization is to replace words with imagery, so if words are still required to make sense of the imagery, heck, why visualize in the first place?
With my own experience in Volume Visualization ( specially Scientific one ) these techniques are just “Colorful” and ( CFD sometimes is jokingly termed as Colorful Fluid Dynamics ( they show only colors) or Congressional Fluid Dynamics ( they give research funds) ) but very rarely they explain the natural phenomenon. The national labs with Petascale initiatives have too many challenges in data visualization and this is one example where visualization doesn’t provide deep insights.