| compute_dic {bvhar} | R Documentation |
Deviance Information Criterion of Multivariate Time Series Model
Description
Compute DIC of BVAR and BVHAR.
Usage
compute_dic(object, ...)
## S3 method for class 'bvarmn'
compute_dic(object, n_iter = 100L, ...)
Arguments
object |
Model fit |
... |
not used |
n_iter |
Number to sample |
Details
Deviance information criteria (DIC) is
- 2 \log p(y \mid \hat\theta_{bayes}) + 2 p_{DIC}
where p_{DIC} is the effective number of parameters defined by
p_{DIC} = 2 ( \log p(y \mid \hat\theta_{bayes}) - E_{post} \log p(y \mid \theta) )
Random sampling from posterior distribution gives its computation, \theta_i \sim \theta \mid y, i = 1, \ldots, M
p_{DIC}^{computed} = 2 ( \log p(y \mid \hat\theta_{bayes}) - \frac{1}{M} \sum_i \log p(y \mid \theta_i) )
Value
DIC value.
References
Gelman, A., Carlin, J. B., Stern, H. S., & Rubin, D. B. (2013). Bayesian data analysis. Chapman and Hall/CRC.
Spiegelhalter, D.J., Best, N.G., Carlin, B.P. and Van Der Linde, A. (2002). Bayesian measures of model complexity and fit. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 64: 583-639.