R: Access model components.

model components {DirichletMultinomial}

R Documentation

Access model components.

Description

The accessors mixture and mixturewt return information about the esimtated Dirichlet comonents of the fitted model. mixture returns a sample x component matrix of estimated values, mixturewt reta matrix of

Usage

mixture(object, ..., assign=FALSE)
mixturewt(object, ...)
goodnessOfFit(object, ...)
laplace(object, ...)
## S4 method for signature 'DMN'
AIC(object, ..., k = 2)
## S4 method for signature 'DMN'
BIC(object, ...)

## S4 method for signature 'DMN'
fitted(object, ..., scale=FALSE)
## S4 method for signature 'DMN'
predict(object, newdata, ..., logevidence=FALSE)
## S4 method for signature 'DMNGroup'
fitted(object, ...)
## S4 method for signature 'DMNGroup'
predict(object, newdata, ..., assign=FALSE)
## S4 method for signature 'DMNGroup'
summary(object, ...)

Arguments

`object`	An instance of class `dmn`.
`newdata`	A `matrix` of new sample x taxon data to be fitted to the model of `object`.
`...`	Additional arguments, available to methods, when applicable.
`assign`	`logical(1)` indicating whether the maximum per-sample mixture component should be returned (`assign=FALSE`), or the full mixture matrix (`assign=TRUE`).
`scale`	`logical(1)` indicating whether fitted values should be returned unscaled (default, `scaled=FALSE`) or scaled by the variability of `mixturewt` parameter `theta`.
`logevidence`	`logical(1)` indicating whether posterior probability (default, `logevidence=FALSE`) or log evidence `logical=TRUE` should be returned.
`k`	ignored.

Value

mixture with assign=FALSE returns a matrix of sample x Dirichlet component estimates. With assign=TRUE mixture returns a named vector indexing the maximal Dirichlet component of each sample.

mixturewt returns a matrix with rows corresponding to mixture components, and columns pi (component weight) and theta (component variability). Small values of theta correspond to highly variable components.

goodnessOfFit returns a named numeric vector of measures of goodness of fit.

laplace, AIC, and BIC return the corresponding measures of goodness of fit.

Author(s)

Martin Morgan mailto:mtmorgan@fhcrc.org

Examples

data(fit)
best <- fit[[4]]
mixturewt(best)
head(mixture(best), 3)
head(mixture(best, assign=TRUE), 3)
goodnessOfFit(best)

fl <- system.file(package="DirichletMultinomial", "extdata",
                  "Twins.csv")
count <- t(as.matrix(read.csv(fl, row.names=1)))
data(bestgrp)
bestgrp
head(predict(bestgrp, count))

[Package DirichletMultinomial version 1.22.0 Index]