compareModels: compute approximate Bayes Factor in favor of one spatial SEIR...

In grantbrown/ABSEIR: Posterior Approximation to Fully Bayesian Spatial SEIR Models Using Multithreaded C++

Description

compute approximate Bayes Factor in favor of one spatial SEIR model over another

Usage

compareModels(
  modelList,
  priors = NA,
  n_samples = 1000,
  batch_size = 10000,
  max_itrs = 1000,
  epsilon = NA,
  verbose = FALSE
)

Arguments

modelList	A list of models to compare with approximate Bayes factors. function.
priors	The prior probabilities of each model in modelList
n_samples	The desired number of accepted simulation values on which the Bayes Factor calculation is to be based
batch_size	The number of epidemics to simulate in parallel before assessing the number of accepted samples
max_itrs	The maximum number of parallel batches to execute before giving up
epsilon	The cutoff value used to determine whether simulated epidemics are accepted or rejected. If left blank, the mean of the two smallest terminating epsilon values models under comparison is used. If these are dramatically different, this approach may produce misleading results.
verbose	A logical value, indicating whether progress information should be displayed.

Details

A Bayes Factor is a measure of the posterior evidence in favor of one model compared to another. In the ABC setting, we may compute approximate Bayes Factors of comparably converged models by assessing the parameter acceptance rate at a new iteration.

Examples

## Not run: compareModels(list(model1, model2))
                                               

grantbrown/ABSEIR documentation built on Oct. 14, 2021, 2:32 p.m.