RPMM (version 1.20)

blcInitializeSplitEigen: Initialize Gaussian Latent Class via Eigendecomposition

Description

Creates a function for initializing latent class model based on Eigendecomposition

Usage

blcInitializeSplitEigen(eigendim = 1,  assignmentf = function(s) (rank(s) - 0.5)/length(s))

Arguments

eigendim

How many eigenvalues to use

assignmentf

assignment function for transforming eigenvector to weight

Value

A function f(x) (see Details.)

Details

Creates a function f(x) that will take a data matrix x and initialize a weight matrix for a two-class latent class model. Here, the initialized classes will be based on eigendecomposition of the variance of x. See blcTree for example of using “blcSplitCriterion...” to control split.

See Also

blcInitializeSplitDichotomizeUsingMean, glcInitializeSplitFanny, glcInitializeSplitHClust