Prev: trojan in 2008b?
Next: Zero-Crossings
From: Bryan on 16 Jun 2010 12:51
I am having trouble figuring out how to display an approximation of
the distance between points in a multi-dimensional space as a 2D
scatter plot, and was hoping someone here could help me out.

I'm not sure if that the above is totally clear, so let me explain. I
have a data set in which 36 objects have been characterized using 27
criteria. As such I have a data table where each row corresponds to
one object, and each column corresponds to that objects score for a
respective criteria.

These scores can be thought of as "positions" in a 27-dimensional
space. Using the pdist command I can calculate the distance between
each of the 36 objects. That distance data can then be used for
cluster analysis and so forth.

But what I want to do is display the 36 objects as a scatter plot;
essentially positioning them such that the distances separating the
objects positions in the 2D scatter plot approximates their "true"
distance from each other in 27-dimensional space. I cannot figure out
how to do this, and I forget the name of the algorithms that do this
kind of approximation.

Any help would be greatly appreciated.

Thanx

Bryan
 | 
Pages: 1
Prev: trojan in 2008b?
Next: Zero-Crossings