MDS_INIT

Initialisation for MDS (variants of Sammon) mapping

    Y = MDS_INIT (D,N,INIT)

Input
D	Square dissimilarity matrix of the size M x M
N	Desired output dimensionality (optional; default: 2)
INIT	Initialisation method (optional; default: 'randnp')

Output
Y	Initial configuration for the MDS method

Description

Finds a configuration of points Y in an N-dimensional space, used as  a starting configuration for an MDS mapping based on the distance matrix D.  The parameter INIT is a string standing for the initialisation method

    'randp'  - linear mapping of D on n randomly (uniform distribution) chosen vectors
    'randnp' - linear mapping of D on n randomly (normal distribution) chosen vectors
    'randv'  - randomly (uniform distribution) chosen vectors from D
    'maxv'   - n columns of D with the largest variances
    'kl'     - Karhunen Loeve projection (linear mapping) of D (first n eigenvectors)
    'cs'     - Classical Scaling

See also

mappings, mds_cs, mds,