SpeciesMix (version 0.3.1)

SpeciesMix.em.gaussian: Implementation of SpeciesMix using EM algorithim

Description

SpeciesMix.em is used to fit mixtures of glms to multivariate species data. The function is a wrapper function for fitMix but allows for multiple EM starts to find the global maxima.

Do not use this function under normal conditions. Documentation is included to allow direct interface with the EM steps.

Usage

#SpeciesMix.em.gaussian(sp.form,sp.data,covar.data,G=2,\n
#em.refit=1,ite.max=500, est.var = FALSE,\n
#residuals=FALSE,trace=TRUE)

Arguments

sp.form

an object of class "formula" (or one that can be coerced to that class):a symbolic description of the model to be fitted

sp.data

a data frame containing the species information. The frame is arranged so that each row is a site and each column is a species. Species names should be included as column names otherwise numbers from 1:S are assigned.

covar.data

a data frame containng the covariate data for each site. Names of columns must match that given in

G

number of archetype species to fit

em.refit

refits model so that the global maxima can be found using EM.

ite.max

maximum number of EM steps to preform

est.var

calculate variance covariance matrix

residuals

calculate residuals

trace

the trace of the EM steps

Value

• logl
  loglikelihood
• aic
  AIC
• bic
  BIC
• ICL
  Integrated Classification Likelihood Criterion
• tau
  posterior probabilities giving group membership
• coef
  coefficents for each species archetype
• fmM
  list of glm objects for archetypes 1 to G
• model.tau
  full matrix of taus used by glm for model fitting. May be useful for residuals

Authors

Piers Dunstan and Scott Foster

Details

SpeciesMix is a wrapper for fitMix that allows for easier data input. The data frames are merged into the appropriate format for the use in fitMix. Multiple models can be fitted so that

Examples

G <-4
S <- 20
theta <- matrix(c(-0.9,-0.6,0.5,1,-0.9,1,0.9,-0.9),4,2,byrow=TRUE)
dat <- data.frame(y=rep(1,100),x=runif(100,0,2.5),z=rnorm(100,10,2))
dat1 <- artificial.data(y~1+x,dat,theta,S)