normalmixMEMtestSeq: normalmixMEMtestSeq

In hkasahar/normalregMix: normalregMix: An R Package for Normal Mixture Regression Models

Description

Sequentially performs MEM test given the data for y and x on the null hypothesis H_0: m = m_0 where m_0 is in 1, 2, ..., maxm

Usage

normalmixMEMtestSeq(y, x = NULL, z = NULL, maxm = 3, ninits = 10,
  maxit = 2000, nbtsp = 199, parallel = 0.75, cl = NULL,
  crit.bootstrap.from = 3)

Arguments

y	n by 1 vector of data for y
x	n by q matrix of data for x (if exists)
z	n by p matrix of regressor associated with gamma
maxm	The maximum number of components set as null hypothesis in the mixture
ninits	The number of randomly drawn initial values.
maxit	The maximum number of iterations.
nbtsp	The number of bootstrap observations; by default, it is set to be 199
parallel	Determines what percentage of available cores are used, represented by a double in [0,1]. 0.75 is default.
cl	Cluster used for parallelization; if it is NULL, the system will automatically create a new one for computation accordingly.
crit.bootstrap.from	The minimum m in null hypothesis to have critical values calculated from bootstrap for the test statistics

Value

A list of with the following items:

alpha	maxm by maxm matrix, whose i-th column is a vector of alphas estimated given the null hypothesis m_0 = i
mu	maxm by maxm matrix, whose i-th column is a vector of mus estimated given the null hypothesis m_0 = i
sigma	maxm by maxm matrix, whose i-th column is a vector of sigmas estimated given the null hypothesis m_0 = i
beta	A list of length maxm, whose i-th element is a q times i matrix of betas estimated given the null hypothesis m_0 = i
gam	maxm by maxm matrix, whose i-th column is a vector of gammas estimated given the null hypothesis m_0 = i
emstat	A maxm vector of values of modified EM statistics of the model at m_0 = 1, 2, ..., maxm
pvals	A maxm by 3 matrix whose i-th row indicates a vector of p-values at k = 1, 2, 3
aic	A maxm vector of Akaike Information Criterion of the fitted model at m_0 = 1, 2, ..., maxm
bic	A maxm vector of Bayesian Information Criterion of the fitted model at m_0 = 1, 2, ..., maxm
loglik	A maxm vector of log-likelihood values of the model at m_0 = 1, 2, ..., maxm
penloglik	A maxm vector of penalized log-likelihood values of the model at m_0 = 1, 2, ..., maxm
pmle.result	A list of output from normalmixPMLE under the number of components selected by sequantial hypothesis testing

Examples

data(faithful)
attach(faithful)
normalmixMEMtestSeq(y = eruptions)

hkasahar/normalregMix documentation built on May 17, 2019, 4 p.m.