Summarize a multiggm_fit object
summary.multiggm_fit.RdDisplays MCMC diagnostics including acceptance rates, edge counts at a given PIP threshold, and graph similarity (theta) estimates.
Usage
# S3 method for class 'multiggm_fit'
summary(object, pip_threshold = 0.5, ...)Arguments
- object
A
multiggm_fitobject returned bymultiggm_mcmc.- pip_threshold
Numeric threshold for counting selected edges via posterior inclusion probability (PIP). Default 0.5.
- ...
Ignored.
Value
An object of class "summary.multiggm_fit" (printed
invisibly) with components:
KInteger; number of groups.
pInteger; number of variables.
nsaveInteger; number of saved posterior draws.
ar_gammaNumeric; mean acceptance rate for between-model (spike-slab toggle) moves on theta.
ar_thetaNumeric; mean acceptance rate for within-model (slab-to-slab) moves on theta.
ar_nuNumeric; mean acceptance rate for nu updates.
edge_countsInteger vector of length K; number of selected edges per group at the given PIP threshold.
pip_thresholdThe PIP threshold used.
theta_meanNumeric matrix
[K, K]; posterior mean of theta (graph similarity) for each pair of groups. Upper triangle only.theta_nonzero_fracNumeric matrix
[K, K]; posterior probability that theta > 0 for each pair. Upper triangle only.hyperNamed list of hyperparameters used in the fit.
Examples
sim <- simulate_multiggm(K = 2, p = 8, n = 80, seed = 1)
fit <- multiggm_mcmc(data_list = sim$data_list, burnin = 200, nsave = 100)
summary(fit)
#> multiGGM MCMC Summary
#> =====================
#> Groups (K): 2 | Nodes (p): 8 | Posterior draws: 100
#>
#> Acceptance rates:
#> gamma (edge toggle): 7.7%
#> theta (within-model): 62.7%
#> nu (edge log-odds): 56.0%
#>
#> Selected edges (PIP >= 0.5 ):
#> Group 1 : 8 edges
#> Group 2 : 7 edges
#>
#> Graph similarity (theta):
#> theta[1,2]: mean = 0.493, P(nonzero) = 97.0%
#>
summary(fit, pip_threshold = 0.3)
#> multiGGM MCMC Summary
#> =====================
#> Groups (K): 2 | Nodes (p): 8 | Posterior draws: 100
#>
#> Acceptance rates:
#> gamma (edge toggle): 7.7%
#> theta (within-model): 62.7%
#> nu (edge log-odds): 56.0%
#>
#> Selected edges (PIP >= 0.3 ):
#> Group 1 : 11 edges
#> Group 2 : 10 edges
#>
#> Graph similarity (theta):
#> theta[1,2]: mean = 0.493, P(nonzero) = 97.0%
#>