robust_sparse_precision: Robustly estimate a precision matrix with a given adjacency...

In Tveten/capacc: Collective and point anomalies in cross-correlated data

Description

The function uses the GLASSAO algorithm implemented in the glassoFast package to estimate a precision matrix robustly with a conditional independence structure specified by adj_mat.

Usage

robust_sparse_precision(x, adj_mat, sparse = TRUE)

Arguments

x	An n x p data matrix where each row is an observation vector.
adj_mat	A p x p adjacency matrix. See details.
sparse	A logical specifying whether the returned estimate should be a sparse matrix from the Matrix package or not.

Details

The function adjacency_mat can be used to generate an adjacency matrix from a list containing the neighbours of variable i in list[[i]]. Moreover, the functions banded_neighbours, lattice_neighbours, random_neighbours can be used to generate such lists of neighbours for banded, lattice and random adjacency structures.

Value

A precision matrix.

Examples

x <- simulate_cor()$x
Q <- robust_sparse_precision(x, adjacency_mat(banded_neighbours(2, ncol(x)), sparse = FALSE))

Tveten/capacc documentation built on Sept. 29, 2021, 5:31 a.m.