The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.
quarto.prder,
pryr, and rmarkdown.get_parameter_dims() no longer requires a compiled Stan
model. This leads to a significant performance improvement when applied
to dynamiteformula objects.cmdstanr backend no longer relies
on rstan::read_stan_csv() to construct the fit object.
Instead, the resulting CmdStanMCMC object is used directly.
This should provide a substantial performance improvement in some
instances. For dynamice(), samples from different imputed
datasets are combined using cmdstanr::as_cmdstan_fit()
instead.gaussian_simulation_fit has been
removed to accommodate CRAN package size requirements. The code to
generate the data is still available in the data_raw
directory.type argument of coef() and
plot() has been replaced by types accepting
multiple types simultaneously, similar to as.data.table()
and as.data.frame().plot_betas(), plot_deltas(),
plot_nus(), plot_lambdas() and
plot_psis() have been deprecated and are now provided via
the default plot method by selecting the appropriate
types.plot_type has been added to control what
type of plot will be drawn by the plot() method. The
default value "default" draws the posterior means and
posterior intervals of all parameters. The old functionality of drawing
posterior densities and traceplots is provided by the option
"trace".plot() method has gained the argument
n_params to limit the amount of parameters drawn at once
(per parameter type).predict() and fitted()
for multinomial responses.cumulative
family are now customizable.factor and ordered factor responses
are now supported for categorical and
cumulative families. In addition,
ordered factor columns of data are no longer
converted to factor columns.rstan and cmdstanr can now be used
interchangeably for either backend, such as iter and
iter_samples.sigma_lambda and
tau_psi, prior is now defined on zeta, the sum
of these, as well as on kappa, which is the proportion of
zeta attributable to sigma_lambda.dynamice()
which uses the mice package.predict and fitted functions no longer
permutes the posterior samples when all samples are used i.e. when
n_draws = NULL (default). This also corrects the standard
error estimates of loo(), which were not correct earlier
due to the mixing of chains.thin for loo(),
predict() and fitted() methods.hmc_diagnostics() is now
available.get_code() and
get_data() functions and how they can be used to modify the
generated Stan code and perform variational Bayes inference.obs(c(y, x) ~ x | 1, family = "mvgaussian").obs(., family = "cumulative", link = "probit") and
obs(., family = "cumulative", link = "logit"),
respectively.data.table version 1.15.0 or
higher and the ggforce package.plot method for dynamiteformula
objects. This method draws a directed acyclic graph (DAG) of the model
structure as a snapshot in time with timepoints from the past and the
future equal to the highest-order lag dependency in the model as a
ggplot object. Alternatively, setting the argument
tikz = TRUE returns the DAG as a character
string in TikZ format. See the documentation for more details.fixed(), varying(), and random()
definitions in obs().character type
group variables.dynamite
which can be used to tweak some aspects of the model (no checks on the
compatibility with the post processing are made).cmdstanr
backend to O0, as the O1 is not necessarily
stable in all cases.full_diagnostics to the
print() method which can be used to control the computation
of the ESS and Rhat values. By default, these are now computed only for
the time- and group-invariant parameters (which are also printed).print() method now also warns about possible
divergences, treedepth saturation, and low E-BMFI.predict() code
generation.dynamite() will now retain the original column order of
data in all circumstances.tau parameters.mcmc_diagnostics() function so that HMC
diagnostics are checked also for models run with the
cmdstanr backend.get_data() method for dynamitefit
objects now correctly uses the previously defined priors instead of the
default ones.data.table package to 1 in examples, tests, and vignettes
for CRAN.lfo() method now uses a single chain and
core to avoid a compatibility issue with CRAN.plot_nus() for categorical responses.predict() and fitted() methods when
newdata contained duplicate time points within group.lfo() in case of missing data.formula.dynamitefit() with models
defined using lags() with a vector k argument
with more than one value.lfo() method which resulted wrong
ELPD estimates in panel data setting.lfo() method which in case of
lagged responses caused the ELPD computations to skip last time
points.dynamite() data parsing that caused
substantial memory usage in some instances.formula.dynamitefit() with models
that had multinomial channels.formula.dynamitefit() when the
df argument of splines() was
NULL.trials() and offset() terms
are now properly parsed when using lags().dynamite() now supports parallel computation via the
reduce-sum functionality of Stan.predict() that resulted in redundant
NAs produced warnings.formula.dynamitefit() with models
that had multivariate channels.update() method used by lfo().update() method for model fit
objects without a group variable.update() method in
lfo()."tau" and "tau_alpha"
type parameters with the as_draws() method for categorical
responses.formula.dynamitefit() when the
model contained a splines component."dynamitefit" objects no longer contain the data used
for Stan sampling by default. This data can still be retrieved via
get_data().gaussian_simulation_fit that
includes the model fit of the dynamite_simulation vignette
for the example with time-varying effects.latent_factor_example and
latent_factor_example_fit have been removed to accommodate
CRAN package size requirements. The code to generate these data is still
available in the data_raw directory.formula.dynamitefit() when the
model formula contained a lags component or a
lfactor component."student" family in obs()."multinomial" family in obs(). A
trials() term is now mandatory for multinomial
channels.trials() and
offset() is now properly checked in the data.trials() and offset()
now function correctly in predict() when they contain
response variables of the model.nobs() for models that have multivariate channels.predict() with models that contained
multivariate channels with random effects.rstan::sampling() and the sample() method of
the cmdstanr Stan model via ... in the call to
dynamite are now checked and unrecognized arguments will be
ignored.get_parameter_dims() that returns
the parameter dimensions of the Stan model for
"dynamitefit" and "dynamiteformula"
objects.noncentered_lambda from
lfactor() as this did not work as intended."nocb".omega parameters, they now include
also the channel name.gregexec() internally which
made it dependent on R version 4.1.0 or higher."mvgaussian" family in obs(). See the
documentation of the dynamiteformula() function for details
on how to define multivariate channels."dynamitefit" class. You can use the
functions get_priors() and
get_parameter_names() to see the names that are available,
as before.verbose_stan is now ignored when
backend = "cmdstanr".stanc_options argument for defining compiler
options when using cmdstanr can now be controlled via
dynamite()."data.table" objects in
predict() leading to faster computation.update() method now checks if the
backend has changed from the original model fit.update() method now properly recompiles the model
(if necessary) in cases where update() is used for already
updated "dynamitefit" object.-Inf prior mean if
all observations at the first time point were zero.plot_deltas() and other plotting functions now throw an
error if the user tries to plot parameters of an incorrect type with
them.dynamite() now supports general group-level random
effects. New random() works analogously with
varying() inside obs(), and the new optional
random_spec() component can be used to define whether the
random effects should be correlated or not and whether to use
noncentered parameterization.bayesplot package.
Instead, ggplot2 and patchwork packages are
used for the plot method.dynamite() function has been
changed: time now precedes group and
backend now precedes verbose. This change is
also reflected in the get_data(),
get_priors(), and get_code() functions.y ~ x and x ~ z simultaneously is
valid, but adding z ~ y to these would result in a
cycle.mcmc_diagnostics() is now clearer.summary argument was changed
to FALSE in as.data.frame() and
as.data.table() methods, whereas it is now hard-coded to
TRUE in the summary() method. The column
ordering of the output of these methods was also changed so that the
estimate columns are placed before the extra columns such as
time.parameters to
as.data.frame() and similar methods as well for the
plotting functions.get_parameter_types() and
get_parameter_names() for extracting model parameter types
and names respectively.data.table package.multichannel_example and the corresponding fit was
modified: The standard deviation parameter of the Gaussian channel used
in the data generation was decreased in order to make the example in the
vignette more interesting.random() component in order to reduce the size of the model
fit object.plot_deltas() no longer unnecessarily warns about
missing values.get_prior(), get_code(), and
get_data() now support case without group
argument, as per issue #48.predict().cmdstanr via argument
backend in dynamite.rstan.dplyr and tidyr to ‘Suggests’.categorical_logit() is now used instead of
categorical_logit_glm() on older rstan and
cmdstanr versions.random() now also support
centered parametrization.formula.dynamitefit() so that it is
now compatible with the update() method. Also added the
required "call" object to the "dynamitefit"
object.loo() and lfo() methods for the
dynamite models which can be used for approximate leave-one-out and
leave-future-out cross validation.env argument of data.table() is now
used to avoid possible variable name conflicts.lambda is now xi in order to free
lambda for factor loadings parameter as is customary in
factor analysis.get_code() applied to fitted model now correctly
returns only the model code and not the stanmodel
object..draw column of the
as.data.frame() output.predict() and
fitted() by separating the simulated values from the
predictors that are independent of the posterior draws.funs, this can further significantly reduce memory usage
when individual level predictions are not of interest.dynamiteThese binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.