brms:::update.brmsfit()
is used with a formula
that was extended on its right-hand side. In those cases, shinybrms now
avoids a recompilation by specifying argument newdata
of
brms:::update.brmsfit()
.brms::gen_extreme_value()
family has been removed from the drop-down
list of advanced distributional families, due to brms's issue
paul-buerkner/brms#1345.
(Note that before, the brms::gen_extreme_value()
family was misclassified
under "Continuous outcome on the positive real line": In fact, the support of
this family depends on its shape parameter.)"home"
when launching the
app. This required increasing the required shiny version to 1.7.0.brms::brm()
to the user.offset()
terms without the need to resort to the previous
workaround based on a constant()
prior (see brms's GitHub issue
#923 solved in brms
v2.16.0). This increases the required brms version to at least 2.16.0.brmsfit
object (page "Posterior",
tab "Run Stan", panel "Run Stan"). This required increasing the default size
limit for file uploads (global option shiny.maxRequestSize
; now at 50 MB if
not set by the user).brmsfit
exists (i.e., it is currently presented on page
"Posterior"), then---if possible---a click on the "Run Stan" button causes this
previous brmsfit
to get updated using brms:::update.brmsfit()
, saving the
compilation time. Note that shinybrms automatically takes care of the fact
that brms:::update.brmsfit()
does not recompute the default priors if the
dataset has changed. However, for uploaded brmsfit
s which were not fitted
through shinybrms, this is not possible (and for uploaded brmsfit
s which
were fitted through shinybrms, this is still complicated), so an uploaded
brmsfit
may not be updated. The updating of a previous brmsfit
may be turned
off using the global option shinybrms.allow_upd
(see ?launch_shinybrms
for
details).brms.backend
can be used to control the default backend selected in the "Advanced options" in
the shinybrms app (see ?launch_shinybrms
for details).FALSE
.brms::parnames()
.parallel::detectCores(logical = FALSE)
instead of
parallel::detectCores()
when determining the default number of cores for the
Stan run.brms::conditional_effects()
).rstan::rstan_options("javascript" = FALSE)
fix on all
platforms (not only Windows) if rstan version 2.21.1 or above is used.launch_shinybrms()
.kidiq
: Converted column mom_hs
to factor.rstan::rstan_options("javascript" = FALSE)
at the launch of the shinybrms
app to prevent occasional crashes of the R session when starting the Stan run.roaches
from package rstanarm.rs_shinyviewer
was used for option "browser"
.hexSticker
.pkgdown
.browser
is avoided. Most importantly, option shinybrms.RStudio_browser
was removed and
replaced by the two new options shinybrms.prog_browser
and
shinybrms.shinystan_browser
.launch.browser
from shinybrms::launch_shinybrms()
so that
the default from shiny::runApp()
is used.adapt_delta
per default to 0.95 as done
in package rstanarm. Set max_treedepth
per default to 15 as done for most
models in rstanarm."Under the hood", there have been many changes which should make the shinybrms app more stable. In particular, the following bugs have been fixed:
First release. Offers support for Bayesian regression models with a Gaussian,
Bernoulli, or negative binomial distribution for the (univariate) outcome. For
the predictors, only nonvarying (a.k.a. population-level or "fixed") effects are
supported. Varying (a.k.a. group-level or "random") effects are not supported
yet. Neither supported are most of
brms's other features, like
monotonic effects for ordinal predictors or non-linear effects. Interactions are
supported, though. For the inspection of the output, only a short summary (from
brms::summary.brmsfit()
) and the possibility to launch
shinystan is offered.
Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.