Difference between revisions of "Multidimensional Scaling"
From Cohen Courses
Jump to navigationJump to search| (One intermediate revision by the same user not shown) | |||
| Line 1: | Line 1: | ||
| − | Multidimensional scaling (MDS) takes input a <math> n \times n </math> matrix of distances <math> D </math> where <math> D_{ij} </math> denotes the target distance between entity <math> i </math> and entity <math> j </math>. It produces an <math> n \times p </math> matrix <math> X </math> where the <math> i</math>th row is the position in ''p''-dimensional latent space. MDS transforms the pairwise distance matrix D into a similarity matrix \tilde D using linear transformations | + | Multidimensional scaling (MDS) takes input a <math> n \times n </math> matrix of distances <math> D </math> where <math> D_{ij} </math> denotes the target distance between entity <math> i </math> and entity <math> j </math>. It produces an <math> n \times p </math> matrix <math> X </math> where the <math> i</math>th row is the position in ''p''-dimensional latent space. MDS transforms the pairwise distance matrix <math> D </math> into a similarity matrix <math> \tilde D </math> using linear transformations. |
[http://en.wikipedia.org/wiki/Multidimensional_scaling External Link] | [http://en.wikipedia.org/wiki/Multidimensional_scaling External Link] | ||
Latest revision as of 16:56, 1 April 2011
Multidimensional scaling (MDS) takes input a matrix of distances where denotes the target distance between entity and entity . It produces an matrix where the th row is the position in p-dimensional latent space. MDS transforms the pairwise distance matrix into a similarity matrix using linear transformations.
