Extract posterior mean partial correlation matrices

Returns a list of K posterior mean partial correlation matrices from a fitted multi-GGM model. This is the fitted S3 method for multiggm_fit objects.

Usage

# S3 method for class 'multiggm_fit'
fitted(object, ...)

Arguments

object

A multiggm_fit object returned by multiggm_mcmc.

...

Ignored.

Value

A named list of K numeric matrices (each p x p). Each matrix is the element-wise posterior mean of the partial correlation matrix \(P_k\) across all saved MCMC iterations, where \(P_{ij}^{(k)} = -\Omega_{ij}^{(k)} / \sqrt{\Omega_{ii}^{(k)} \Omega_{jj}^{(k)}}\). Diagonal entries are 1. List names are "Group_1", "Group_2", etc.

Details

For each saved iteration, the precision matrix \(\Omega_k\) is converted to a partial correlation matrix, then the posterior mean is taken element-wise. This is computed via posterior_pcor.

See also

coef.multiggm_fit(), posterior_pcor(), precision_to_pcor()

Examples

sim <- simulate_multiggm(K = 2, p = 8, n = 80, seed = 1)
fit <- multiggm_mcmc(data_list = sim$data_list, burnin = 200, nsave = 100)
pcor_hat <- fitted(fit)
round(pcor_hat$Group_1[1:5, 1:5], 3)
#>        [,1]   [,2]   [,3]   [,4]   [,5]
#> [1,]  1.000 -0.480 -0.014  0.005 -0.003
#> [2,] -0.480  1.000 -0.020  0.005  0.050
#> [3,] -0.014 -0.020  1.000  0.119 -0.097
#> [4,]  0.005  0.005  0.119  1.000 -0.219
#> [5,] -0.003  0.050 -0.097 -0.219  1.000