init {UEM}R Documentation

Initializing the EM algorithm

Description

Functions to initialize the EM algorithm.

Usage

init(y, K, type = "quantile", tol=0.5,  ...)
shortruns(y, K, init = "random", family = "Gaussian", maxit = 50, 
    threshold = 10, lambda = 1, verbose = TRUE, plot = FALSE)

Arguments

y

a univariate or multivariate data set.

K

the number of components.

type

type of starting points. Includes "random" (randomly chosen from the data), "scatter" (uniformly sampled from the _support_ of the data), "quantile" (quantile-based), "shortruns" (short runs of EM), "gq" (based on Gauss Quadrature points).

tol

only relevant for type="gq". Scales Gaussian Quadrature points inwards or outwards by the factor tol.

init
family
maxit
threshold
lambda
verbose
plot
...

Author(s)

J. Einbeck and D. Bonetti


[Package UEM version 0.3-1 Index]