SIN (version 0.6)

sinCG: SIN for chain graphs

Description

This function computes the matrix of simultaneous p-values for SIN model selection for chain graphs. SIN assumes that a dependence chain or blocking of the variables is available.

Usage

sinCG(blocks, S, n, type="AMP", holm=TRUE)

Arguments

blocks

a list of integer vectors with entries amongst $1,\ldots,p$ where $p$ is the number of variables. Each one of the integer vectors specifies a set of variables that form a block in the chain graph. Furthermore, a partial ordering of the variables is specified by the convention that variables in one block are ordered smaller than variables in a block succeeding in the list blocks.

S

a covariance or correlation matrix.

n

the sample size.

type

a string equal to either "AMP" or "LWF" that determines whether the chain graph is interpreted with respect to the Andersson, Madigan, Perlman Markov property or the Lauritzen, Wermuth, Frydenberg Markov property.

holm

Boolean variable indicating whether Holm's p-value adjustment should be used (holm=TRUE) or not (holm=FALSE).

Value

A matrix of simultaneous p-values with NA on the diagonal.

References

Drton, M. \& Perlman, M.D. (2004) Model Selection for Gaussian Concentration Graphs. Biometrika 91(3): 591-602. Drton, M. \& Perlman, M.D. (2008) A SINful Approach to Gaussian Graphical Model Selection. J. Statist. Plann. Inference 138(4): 1179-1200. Andersson, S.A., Madigan, D. \& Perlman, M.D. (2001) Alternative Markov Properties for Chain Graphs. Scandinavian Journal of Statistics 28(1): 33-85. Lauritzen, S. (1996) Graphical Models. Oxford University Press: Oxford.

See Also

plotCGpvalues

Examples

data(fowlbones)
p <- dim(fowlbones$corr)[1]
blocks <- list(1:2,3:4,5:6)
sinCG(blocks,fowlbones$corr,fowlbones$n, type="AMP")
sinCG(blocks,fowlbones$corr,fowlbones$n, type="LWF")
sinCG(blocks,fowlbones$corr,fowlbones$n, type="AMP", holm=FALSE)
sinCG(blocks,fowlbones$corr,fowlbones$n, type="LWF", holm=FALSE)