-3

I want to represent 4 dimensional data in RGB colors so that when clustering is done similar nodes have similar colors or by position on an XY grid. how can this be done?

iteratorr
  • 149
  • 1
  • 8
  • If you're asking what I think you're asking, then RGB isn't typically a good choice because adding 2 RGB colours might not work the way you expect it to. I find the HSV colour space is easier to comprehend and produces more meaningful results when you are attempting to visualize data. http://www.colourlovers.com/blog/2008/08/13/demystifying-rgb-vs-hsv/ You may need to be a little more illustrative of your data if you want a decent answer though. Your question is a bit vague. – Robert Paulson Jun 15 '10 at 22:18

2 Answers2

0

4d? like 3d data with the time dimension?

Anyway, it depends on what technology we're talking about. It sounds like you would need a 3d scatter chart with your fourth dimension being a selection control such as a slider that changes the time. You could use WPF w/ silverlight if you want to display it on a site but I'm sure a variety of technologies can be used to get the same result. You'll need to modify most common solutions to get the similar colors according to position. Statistical algorithms like k-means is usually suited for clustering data together.

James Santiago
  • 2,883
  • 4
  • 22
  • 29
  • 3d scatter is good except I want this to produce graphs for a paper and 3d graphs don't really make good representations on a 2d statis surface, I was hoping I could get just convert 4d info to 3d while keeping their relative values correct – iteratorr Jun 16 '10 at 05:03
  • I don't see how this is a programming question. Can you explain the data types of each dimension? Can you explain what technology you plan on using to implement this data display? How about the ranges of each dimension. When I think of 4d data, I think of something like the position of stars over a period of time(1d = x, 2d = y, 3d = z, 4d = time). – James Santiago Jun 16 '10 at 05:53
-1

Let the four dimensions be w,x,y,z.

There will only be 3 axis:

  • make x as the no. of degrees it is farther from x axis
  • similarly, put the value of y as the no. of degrees it is parallel to y axis.
  • do the same with z.

w represents how far the point is from the origin of the chart.

Alternatively, you may simply use a Radar Chart.

Alex
  • 23,004
  • 4
  • 39
  • 73
Shreyas
  • 1