loglikelihood_global_admkr

loglikelihood_local_admkr

resid


The log marginal likelihood can be computed as the log likelihood + log prior - log posterior. The latter one can be estimated from the MCMC.


method

Estimating optimal bandwidths for the regression mean function approximated by the functional Nadaraya-Watson estimator and the error density approximated by a kernel density of residuals simultaneously in a scalar-on-function regression. As a by-product of Markov chain Monte Carlo, the optimal choice of semi-metric is selected based on largest marginal likelihood.

loglikelihood_global_admkr: Compute the marginal likelihood using Chib's (1995) method

Description

Usage

Arguments

Value

References

See Also