Comments on: Design Prototype

By: dhe

dhe — Thu, 11 Mar 2010 15:45:57 +0000

The use of symmetry for displaying correlation is interesting. Could this lead to comparison of multiple frames through the tiling of triangles? What would this look like in a kaleidoscope?

By: Nate

Nate — Thu, 11 Mar 2010 15:12:12 +0000

The cylinder vis is an interesting idea; however, it seems to be particularly tricky for me to judge either precise angles in 3D, or precise colors — so it’s hard to tell how close two lines are, especially when on opposite sides of the cylinder.

The more I think about the symmetry view, however, the more I like it. True, the exact details of the design will need to be chosen to to help weed out false symmetries (maybe an “overlay” view would highlight those?), but this is a neat idea that uses both our skill at detecting symmetries, and the fact that the data in these networks should be at least somewhat spatially correlated.

By: watkins

watkins — Thu, 11 Mar 2010 11:16:12 +0000

Regarding the first visualization, I agree with Danielle– a 3D visualization might create some problems. It seems like the curve of the cylinder will affect people’s ability to accurately gauge distances between points, especially closer to the sides. However, I like the idea of being able to focus on either low-correlation or high-correlation data separately.

The second visualization is a really cool idea. I like that you used a dataset where it shows information well and one where it wasn’t as clear. That seems like a great way to help pinpoint specific problems, and help everyone improve their designs.

By: Jeremy White

Jeremy White — Thu, 11 Mar 2010 08:18:03 +0000

Treating the ‘symmetric’ designs as two separate entities might be useful. A flipbook could be used to show the different matrices as overlays with the same orientation, instead of mirroring the graphics along the same axis. This would allow a direct comparison to be made without the need to mentally reorient the graphics.

By: Jim Hill

Jim Hill — Thu, 11 Mar 2010 07:22:39 +0000

The first attempt was probably the most imaginative in the class. I would worry about the random positioning of each data point on either side of the zero point. It would also be nice to see this with more than one frame.

The second attempt really made the frame data pop out. I think there’s a lot of promise there.

By: dalbers

dalbers — Thu, 11 Mar 2010 07:09:50 +0000

The second plot is really nice as far as comparing two matricies side-by-side. Pushing two halves of the symmetric matricies together nicely exploits the nature of the data; although the false pattern matchings as pointed out in class certainly are a perceptual problem.

The first visualization looks really nice and is very different from the other proposed solutions. However, managing 3D visualizations is well documented as a visual problem. Between the 3D problems and the complexity of the visualization, it may be very difficult for the user to determine any significant trends in the data.

By: lyalex

lyalex — Thu, 11 Mar 2010 06:36:41 +0000

I’m a little bit confused by the cylinder design. For me, since each matrix has its own distance towards any other matrix, there might be specified cylinders for each one of them (if we define the distance of two matrix by some way, distance of matrix 3 and matrix 1, “D31” is not necessarily equal to “D32+D21”) thus, what’s the superity for the cylinder design?

For plot 2, the different between this method and some others is that this one actually put two matrice side by side to compare, while others are overlapping them.

By: Adrian Mayorga

Adrian Mayorga — Thu, 11 Mar 2010 03:04:35 +0000

I like the fact that you guys thought about the problem at different scales. However like you pointed out, even if you notice a slight difference in the higher symmetry comparison, its hard to tell exactly what it going on. Maybe if you linked the two visuals somehow, you could have a sort of hierarchical exploration system.

Also, how would the symetric based visualization scale when more than two frames need to be compared?

By: gleicher

gleicher — Mon, 08 Mar 2010 03:43:11 +0000

i am really intrigued by the “symmetry seeking” design – i keep wanting to look at things side by side (rather than mirrored), but you would think that symmetry detection might be perceptually more significant.

By: punkish

punkish — Sat, 06 Mar 2010 17:04:17 +0000

The first example, the rotating cylinder dingus, is a really different departure from the usual visualizations. It requires a bit of work still to make the overall interpretation more easy and obvious, but my kudos to the team for thinking outside the square.

The second example is also a very nice example of lateral thinking… throwing away a symmetrical triangle of data and replacing it with a triangle of data from a different dataset.

Very nice job overall.