ggm_search: Gaussian graphical model: automated search

Description Usage Arguments Details Value Note References Examples

View source: R/ggm_search.R

Description

Data mining to learn the graph.

Usage

1
2
3
4
5
6
7
ggm_search(
  x,
  IC = "BIC",
  type = "neighborhood_selection",
  method = "forward",
  n = NULL
)

Arguments

x

A data matrix of dimensions n (observations) by p (nodes) or a correlation matrix of dimensions p by p.

IC

Character string. The desired information criterion. Options include "AIC" and "BIC" (default).

type

Character string. Which search method should be used? The options included "regression" and "approx_L0". See details.

method

Character string. The desired subset selection method Options includes "forward" (default), "backward", and "exhaustive".

n

Integer. Sample size. Required if a correlation matrix is provided.

Details

type = "neighborhood_selection" was described in \insertCitewilliams2019nonregularized;textualGGMnonreg and type = "approx_L0" was described in \insertCitewilliams2020beyond;textualGGMnonreg. The penalty for type = "approx_L0" is called seamless L0 \insertCitedicker2013variableGGMnonreg

Value

An object of class ggm_search, including wadj (weighted adjacency matrix) and adj (adjacency matrix).

Note

type = "neighborhood_selection" employs multiple regression to estimate the graph (requires the data), whereas type = "approx_L0" directly estimates the precision matrix (data or a correlation matrix are acceptable). If data is provided and type = "approx_L0", by default Pearson correlations are used. For another correlation coefficient, provide the desired correlation matrix.

type = "approx_L0" is a continuous approximation to (non-regularized) best subset model selection. This is accomplished by using regularization, but the penalty (approximately) mimics non-regularized estimation.

References

\insertAllCited

Examples

1
2
3
4
5
# data
Y <- ptsd

# search data
fit <- ggm_search(Y)

GGMnonreg documentation built on April 8, 2021, 5:06 p.m.