projsel() if the number of observations in
the dataset is smaller than both the number of available predictors and
the maximum number of iterations in the selection procedure.rstan 2.21.RcppParallel to Imports and LinkingTo, as future
versions of rstan require to link to the Intel TBB
library.sub.idx option to
posterior_performance() to select the observations to be
used in the computation of the performance measures.start.from option to run projsel()
to start the selection procedure from a submodel different from the set
of unpenalized covariates.posterior_linpred()
and projsel(): this also benefits all other functions that
use posterior_linpred(), such as log_lik(),
posterior_predict(), posterior_performance()
and others.posterior_performance() for Windows.posterior_performance() for gaussian
models.projsel() on models with no penalized
predictors.normal_id_glm() and
bernoulli_logit_glm().iter and warmup options in
kfold().rstantools 2.0.0.slab.scale parameter of
hsstan(), as it was not squared in the computation of the
slab component of the regularized horseshoe prior. The default value of
2 in the current version corresponds to using the value 4 in versions
0.6 and earlier.kfold() and posterior_summary()
functions.parallel::parLapply().sample.stan() and
sample.stan.cv().get.cv.performance() with
posterior_performance().projsel().plot.projsel() for choosing the number
of points to plot and whether to show a point for the null model.mc.cores option when
loading the package.projsel() only if selection
stopped early.max.num.pred argument of
projsel() to max.iters.rstan::sampling().hsstan().hsstan().foreach()/%dopar% with
parallel::mclapply().posterior_interval(),
posterior_linpred(), posterior_predict()
log_lik(), bayes_R2(), loo_R2()
and waic() functions.nsamples() and sampler.stats()
functions.crossprod()/tcrossprod() instead of
matrix multiplications.fit.submodel().log_lik() instead of computing and storing the
log-likelihood in Stan.pars in
summary.hsstan().sample.stan() and sample.stan.cv()
into hsstan().loo() method for hsstan objects.adapt.delta argument for base models
from 0.99 to 0.95.scale.u from 20 to 2.scale() to standardize the data in
sample.stan.cv().plot.projsel().get.cv.performance() work also on a
non-cross-validated hsstan object.print() and summary() functions for
hsstan objects.plot.projsel().adapt_delta parameter and change
the default for all models from 0.95 to 0.99.doParallel since doMC is not
packaged for Windows.plot.projsel().
Need a high-speed mirror for your open-source project?
Contact our mirror admin team at info@clientvps.com.
This archive is provided as a free public service to the community.
Proudly supported by infrastructure from VPSPulse , RxServers , BuyNumber , UnitVPS , OffshoreName and secure payment technology by ArionPay.