dimest.cv: Estimate dimensionality by crossvalidation

In markallenthornton/dimest: Dissimilarity Matrix Dimensionality Estimation

Description

Estimate dimensionality by change in crossvalidated multidimensional scaling fit

Usage

dimest.cv(dmat,epsratio=.001,mds.control=list(type="ordinal",itmax=1000))

Arguments

dmat	m x n matrix of n vectorized dissimilarity matrices epsratio convergence criterion. Defaults to .001 mds.control list of smacofSym parameters: type, defaults to "ordinal"; itmax, defaults to 1000

Details

Defines identity matrix of dimensions n x n. Reproduces functionality of identically named MATLAB function.

Value

ndim estimated number of dimensions cv.cor vector of mean crossvalidated correlations between mds configurations and left out data

See Also

smacofSym dimest dimest.R2

Examples

set.seed(1)
dat <- matrix(rnorm(200),100,2)
dmat <- as.vector(dist(dat))
dmat2 <- replicate(2,dmat+rnorm(4950))
dimcvfit <- dimest.cv(dmat2)

markallenthornton/dimest documentation built on May 21, 2019, 11:48 a.m.