summary.XDNUTS {XDNUTS} | R Documentation |
Function to print the summary of an XDNUTS model.
Description
Function to print the summary of an XDNUTS model.
Usage
## S3 method for class 'XDNUTS'
summary(
object,
...,
q.val = c(0.05, 0.25, 0.5, 0.75, 0.95),
which = NULL,
which_chains = NULL,
digits = 5
)
Arguments
object |
an object of class |
... |
additional arguments to customize the summary. |
q.val |
desired quantiles of the posterior distribution for each coordinate.
Default values are |
which |
either a numerical vector indicating the index of the parameters of interest or a string
where both |
which_chains |
a numerical vector indicating the index of the chains of interest. |
digits |
number of digits in the summary table. |
Value
a list containing a data frame named stats
with the following columns:
- mean
the mean of the posterior distribution.
- sd
the standard deviation of the posterior distribution.
- q.val
the desired quantiles of the posterior distribution.
- ESS
the Effective Sample Size for each marginal distribution.
- R_hat
the Potential Scale Reduction Factor of Gelman (Gelman and Rubin 1992), only if multiple chains are available.
- R_hat_upper_CI
the upper confidence interval for the latter, only if multiple chains are available.
Other quantities returned are:
- Gelman.Test
the value of the multivariate Potential Scale Reduction Factor test (Gelman and Rubin 1992).
- BFMI
the value of the empirical Bayesian Fraction of Information Criteria (Betancourt 2016). A value below 0.2 indicates a bad random walk behavior in the energy Markov Chain, mostly due to a suboptimal specification of the momentum parameters probability density.
References
Betancourt M (2016).
“Diagnosing suboptimal cotangent disintegrations in Hamiltonian Monte Carlo.”
arXiv preprint arXiv:1604.00695.
Gelman A, Rubin DB (1992).
“Inference from iterative simulation using multiple sequences.”
Statistical science, 7(4), 457–472.