Tests for convergence in likelihood parameters
Source:R/converge_likelihood.R
converge_likelihood.RdGelman-Rubin convergence tests for each defining parameter in the likelihood distribution.
Arguments
- jags_model_run
The output list returned from
run_likelihood().
Value
A tibble of parameters and convergence results from coda::gelman.diag(),
Values greater than 1.2 indicate problems in convergence. Values between 1.1
and 1.2 indicate weak convergence. Values less than 1.1 indicate good
convergence. This test indicates if the 3 mcmc chains are will mixed and
stable but is insufficient as the only indicator of convergence. Visual plots
of posterior distributions should be investigated as well.