coef() methods and tidy.mcmcr() gain a directional_information argument.
When TRUE the svalue column reports
extras::directional_information()
in place of extras::svalue() (#71, #84).
The default is currently FALSE and will change to TRUE in a future
release; calling either function without setting directional_information
now signals a deprecation warning so the change can be made explicit.tidy.mcmcr() now defaults to simplify = TRUE, matching the nlist
tidy() methods.coef(simplify = FALSE), deprecated in 0.4.1, is now defunct (#74).parameters() and parameters<-(), which are
defunct in term, have been removed (#85).rhat.mcmcrs(bound = TRUE) now returns a named list of scalars rather than a
single scalar; use rhat(x, bound = TRUE)$bound for the previous behavior.
The change is signaled with a warning (#74).The following, soft-deprecated in 0.2.1, now warn on every use (#72):
terms(); use as_term() instead.zero(); use fill_all() instead.check_mcmcarray() and check_mcmcr(); use chk_mcmcarray() and
chk_mcmcr() instead.subset(iterations = ) and subset(parameters = ); use subset(iters = )
and subset(pars = ) instead.pars(terms = ); use term::pars_terms(as_term(x)) for terms = TRUE, and
pars(x) for terms = FALSE.extras is now required at version 0.10.0 or later, and nlist at
version 0.5.0 or later.bound = TRUE rhat() now also returns rhat values for separate analyses.bound = TRUE and as_df = TRUE rhat() now returns a data.frame with the rhat values for the separate and combined analyses.fill_na() for mcarray, mcmcarray and mcmcr.as.mcmcarray.mcmc() (and as.mcmcr.mcmc()) so now returns an mcmcarray (and mcmcr) object with no terms.tidy.mcmcr().simplify = FALSE argument to coef() and tidy() and soft-deprecated if not TRUE.... optional arguments for fun = median argument to estimates().as_nlists.mcmc.list() to nlist package.as_mcmc_list.mcmr().nlist
as_nlist.mcmc() and as_nlist.mcmc.list()as_nlists.mcmc()as.term.mcmc() and as.term.mcmc.list()bind_iterations.mcmc() and bind_iterations.mcmc.list()collapse_chains.default() and collapse_chains.mcmc.list()Changed
npdims.mcmc.list() to return character vector (as opposed to list)collapse_chains.mcmc.list() to return an mcmc.list object with one chain (as opposed to an mcmc object)estimates() from object to x.scalar_only = FALSE argument of pars() to scalar = NA.estimates() so now checks fun returns scalar numeric.Soft-deprecated
pvalue() for extras::pvalue().zero() for fill_all().check_mcmcarray() and check_mcmcr() for chk_mcmcarray() and chk_mcmcr().iterations argument with iters in subset().parameters argument with pars in subset().Added
vld_() and chk_() functions for mcmcarray and mcmcr objects.scalar = NULL argument to pars() and npars().na_rm = NA argument to esr() and rhat().as_df = FALSE arg to esr() for mcarray, mcmc and mcmc.list.Moved
nchains(), niters(), collapse_chains() and split_chains() etc to universals package.check_mcmcr() and check_mcmcarray().converged().as.mcmc.mcmc.list(), thin.mcmc() and thin.mcmc.list() as now defined by coda.as.mcmc.list.mcarray() as clashes with rjags version.mcmc_aperm() function to transpose parameter dimensions.npdims() function to get number of parameter dimensions.by = TRUE argument to mcmc_map() function.rhat() now returns minimum of 1.subset() and parameters() for mcmcrs object.bound = FALSE argument to rhat.mcmcrs() and converged.mcmcrs() functions.error() with err::err().