emnr: EM algorithm

In Hlch1992/ARZIMM: Statistical modeling and inference of microbial interaction and stability for microbial dynamical systems

Description

EM algorithm used in the ARZIMM model.

Usage

emnr(data, para, weight, family, selgamma = FALSE)

Arguments

data	a list of data with componenets: yFdata a vector of xFdata a matrix of variables corresponding to the non-zero auto-regressive model conFdata a matrix of covariates corresponding to the zero state logit model group a vector of numbers as group indicator
para	a list of parameter estimates: beta initial value for beta gamma initial value for gamma sigma initial value for sigma
weight	a vector of observation weightsfor both the non-zero auto-regressive model and the zero state logit model
family	a character string indicating the distribtuion. default is Poisson
selgamma	logical; should concomitant variables in the zero state logit model be selected

Value

a list of fits

para	a list of parameter estimates: beta beta estimates gamma gamma estimates sigma sigma estimates
ciestm	the estimated random effects
conv	logical; did the algorithm converged
df	the number of non-zero parameter estimates
bic	a vector of BIC, AIC, and log likelihood
mse	a vector of square root of mean (pearson) standard error
lambda	a vector of lambda sequence

Hlch1992/ARZIMM documentation built on Feb. 11, 2020, 2:34 a.m.