Comments on: Design challenge: one of our designs

By: Nate

Nate — Thu, 11 Mar 2010 15:42:00 +0000

Super clever! Of the “let’s make the matrix into a shape” designs, I like this one quite a lot. The interaction seems a little fiddly — it’s easy to make things spin when you don’t mean to — but other than that, it’s a great way to visually represent the data.

One thing I do wonder: how easy would it be to indicate “classes” of data labels? Say, the “skill / knowledge / identity” concepts?

By: dhe

dhe — Thu, 11 Mar 2010 15:17:50 +0000

The navigation of this visualization is impressive. In both the reference frame selection and reference concept selection, one can iterate forward, iterate backward, and arbitrarily select.

By: Jim Hill

Jim Hill — Thu, 11 Mar 2010 07:17:44 +0000

I thought this was a great visualization. The best idea was the use of circles to show relative distances from the center. I should have thought of that for my solution. My only question is one of scalability. With 20 frames, the selection circle gets pretty big. This could be fixed by moving the circle off to the side.

By: lyalex

lyalex — Thu, 11 Mar 2010 06:45:12 +0000

I think it’s a good design. The “transmitting ray” design give it space to extend for larger sample sets, and the coding translated comparison of data in a row into comparison of octagons. I think the only problem might be that it can not be used to tell the difference of 2 matrices in a macro view, and when the number of the rows grows, it can eventually have clutter problems.

By: dalbers

dalbers — Thu, 11 Mar 2010 06:32:21 +0000

I generally agree with everyone else on this. The visualization definitely has the “cool” factor in it’s presentation. However, because interaction is so heavily weighted, it would be nice to be able to anchor certain views in order to keep track of trends seen thus far and perhaps then build on the exploration given the marked setting of the view.

By: Shuang

Shuang — Thu, 11 Mar 2010 04:54:41 +0000

This radar map is quite nice and I like the color selection. Also, it gives a way to view each variable one by one, which makes the plot neat. I guess the orientation can be changed according to certain setting, and the angel between variables can also be changed to represent their closeness.

By: turetsky

turetsky — Thu, 11 Mar 2010 03:14:11 +0000

I really like this view. The implementation and the interaction is nice. There is little clutter and you can compare between data sets nicely. I would like to see it have an option of not reorienting when choosing a new variable. That way you can simply cycle through the variables and see how everything compares within one frame of reference.

By: ChamanSingh

ChamanSingh — Thu, 11 Mar 2010 01:47:28 +0000

In my view, this was the best representation of dataset. It has all the simplicity, and sparseness. I presume that this will scale quite nicely to reasonably large dimensions.

By: Nakho Kim

Nakho Kim — Sun, 07 Mar 2010 22:49:21 +0000

A great way to browse through radar maps of each variable. It would be great if users could open several windows with different vaiables and compare them at the same time.

By: jeeyoung

jeeyoung — Fri, 05 Mar 2010 23:09:10 +0000

Nicely implemented. The concentric circles might cover the radial patterns in some cases.