Exteriorpts

From Eigenvector Research Documentation Wiki
Revision as of 13:19, 24 September 2010 by imported>Donal (→‎Purpose)
Jump to navigation Jump to search

Purpose

Finds pts on the exterior of a data space after data are normalized.

Synopsis

[loads,isel] = exteriorpts(x,ncomp,options)

Description

The data (X) are assumed to be all >0 and to be modelable as: X = CS' + E

Inputs

  • x = MxN matrix.
  • ncomp = number of components to extract.

Outputs

  • loads = cell array with extracted pts/factors. Non-selectdim modes are determined via projection.
  • isel = cell array with isel{options.selectdim} containing indices of selected pts.

Options

options = a structure array with the following fields:

  • display : [ 'off' | {'on'} ] governs level of display to command window.
  • samplemode : [ 1 ] mode that contains variance (factors for other modes are normalized to unit 2-norm).
  • distmeasure : [ {'Euclidian'} | 'Mahalanobis' ]
  • offset : [ 0.001 ] approximate noise level, points with 1-Norm <minnorm*max(norm(x)) are not used.

See Also

als, mcr, parafac, purity, purityengine