Extract the estimated inverse covariance from an msfit_ggm
object.
Bayesian model averaging is used, optionally entries with posterior
probability of being non-zero below a threshold are set to 0.
Usage
icov(fit, threshold)
Value
Estimated inverse covariance matrix.
Arguments
fit
Object of class msfit_ggm, returned by
modelSelectionGGM
threshold
Entries with posterior probability of being non-zero
below threshold are set to 0. If missing this argument is ignored and
no entries are set to exact zeroes. When the goal is to identify
zeroes, a sensible default is threshold=0.95
Author
David Rossell
Details
The inverse covariance is obtained via Bayesian model averaging, using
posterior samples of Omega. When threshold is specified,
entries in the BMA estimate are set to zero, which may result in a non
positive-definite matrix.
See Also
modelSelectionGGM,
coef.msfit_ggm for Bayesian model averaging estimates and
intervals.