This function is extensible: new html elements can be provided, but at the moment it is not very user-friendly to do so.
make_MCMC_comparison_pages(comparisonResults, dir = ".", pageComponents,
modelNames = names(comparisonResults), control, plot = TRUE)
An object returned by compareMCMCs
(or combine_MCMC_comparison_results
or rename_MCMC_comparison_method
).
Directory in which to generate html output (default is current working directory)
An optional named list indicating which components to include in the html page for each MCMC method. Each element should be TRUE or FALSE. Options include: timing
(default FALSE): a table of computation times, which is not very useful by itself without information about mixing. efficiencySummary
(default FALSE): A summary of minimum and mean MCMC efficiencies (over all top-level parameters), defined as effective sample size / computation time. efficiencySummaryAllParams
(default TRUE): Like efficiencySummary
, but with the addition of a figure that shows the MCMC efficiency for each top-level parameter. efficiencySummaryAllParams
(default TRUE): Like efficiencySummaryAllParams
, but showing MCMC Pace, defined as 1/MCMC Efficiency. efficiencyDetails
(default TRUE) grid of efficiency barplots for each top-level parameter. posteriorSummary
(default TRUE): grid of mean, median, and 95% credible interval for each top-level parameter.
names to use for the models. By default these will be taken from comparisonResults
.
Optional list with elements makeTopPage
and/or mainPageName
. makeTopPage
has to do with whether an html index page should be generated to link to the results from each model in the comparisonResults
. makeTopPage
can take values 'yes', 'no', and 'if_needed', where the last case indicates to make an index page only if there is more than one model in comparisonResults
. mainPageName
gives a name for the html file if it is generated, with default 'main' (to which '.html') is appended.
If TRUE, the html files are actually generated. If FALSE, they are not generated, and instead a list with the figures is returned. Each element of the list can be rendered as the argument of print
or plot
.
See argument plot
Effective sample size (ESS) is calculated using the coda
packages effectiveSize
function. In cases of poot mixing, this can over-estimate ESS by not using a sufficiently high-order auto-regressive model in the estimation procudure. See updateMCMCcomparisonWithHighOrderESS
for other options.