initialize_fun: A function for initialization.

In yuanfang90/LNMVGA: Cluster microbiome / compositional data using a mixture of logistic-normal multinomial models

A function for initialization.

Description

This function performs initialization of the class membership providing the option of using k-means start, ramdom start, and a small EM or gridsearch type of start: initial randomly on a given number of random starting points, and run for a given number of iterations, then pick the one that ends with highest approximated log-likelihood.

Usage

initialize_fun(W, G, K, n, method, nrep = NA, niter = NA, verb = FALSE)

Arguments

W	observation data: the count data, n*(K+1) matrix where K is the number of taxa.
G	number of component.
K	number of taxa.
n	number of observations.
method	string variable, could be "kmeans", "random", and "small_EM".
nrep	number of random starts. Only useful if method = "small_EM" is specified. default is 50.
niter	number of iterations for each small EM random start. Only useful if method = "small_EM" is specified. default is 10.
verb	logical variable. if is TRUE, each iteration of the small_EM start will be printed.

Examples

initialize_fun()

yuanfang90/LNMVGA documentation built on Jan. 29, 2024, 8:24 a.m.