sparsify.fused: Determine support of multiple partial correlation/precision...

In CFWP/rags2ridges: Ridge Estimation of Precision Matrices from High-Dimensional Data

Determine support of multiple partial correlation/precision matrices

Description

A simple wrapper for sparsify which determines the support of a list of partial correlation/precision matrix by various methods and returns the sparsified matrices.

Usage

sparsify.fused(Plist, ...)

Arguments

Plist	A list of numeric precision matrices.
...	Arguments passed to sparsify.

Value

A list of the same length as Plist with the output from sparsify.

Author(s)

Anders Ellern Bilgrau, Wessel N. van Wierigen, Carel F.W. Peeters <[e-mail]>

See Also

sparsify

Examples

ns <- c(10, 11)
Ylist <- createS(ns, p = 16, dataset = TRUE)
Slist <- lapply(Ylist, covML)
Tlist <- default.target.fused(Slist, ns)

# Obtain regularized precision under optimal penalty
opt <- optPenalty.fused.auto(Ylist, Tlist, cv.method = "aLOOCV",
                            maxit.ridgeP.fused = 1500)
# Use the optimal penalties
Plist <- ridgeP.fused(Slist, ns, lambda = opt$lambda, maxit = 1000)

# Determine support regularized (standardized) precision under optimal penalty
res <- sparsify.fused(Plist, threshold = "top", verbose = FALSE)
round(res[[1]]$sparsePrecision, 1)
round(res[[2]]$sparsePrecision, 1)

CFWP/rags2ridges documentation built on Oct. 21, 2023, 10:19 a.m.