sorting points into clusters (list)

Replies to This Discussion

Permalink Reply by Hannes Löschke on April 17, 2012 at 3:21am

Each cluster would have it's own distance threshold, so I'd say iteration is the way.
You'd need to find closest pairs and collapse those into one point. Repeat that until you found the desired number of clusters.

Permalink Reply by Mateusz Zwierzycki on April 17, 2012 at 6:25pm

My idea for that (not sure if it works) - 3d kmeans, where third dimension is distance to closest point.. dimensions should be weighted (x*1, y*1, z*e.g. 0.1) to manage with overlapping convex hulls (probably they wont be convex).

...but in this particular example 2d kmeans (x and y position) should work. search this forum for kmeans (there was a good script for that somewhere)

Permalink Reply by bea franko on April 18, 2012 at 5:07am

hm... i think i dont get it...

could you help with a defininition?

Permalink Reply by Fred Becquelin on April 18, 2012 at 6:14am

Did you see this?

http://www.grasshopper3d.com/forum/topics/kmeans-1

Permalink Reply by bea franko on April 18, 2012 at 7:08am

thank you so much!!!

Permalink Reply by Yoann Mescam (Systemiq) on April 18, 2012 at 6:56am

http://www.grasshopper3d.com/group/coffee-and-grasshopper/forum/top...

Could be useful, some manual settings but the k-mean approach can perhaps complement it.