coda
v0.4 or higher.Rcpp
v0.11.0 or higher.tibble
v2.1.3 or higher.teacher_rate
for demonstration in vignette.prep_docs()
to prepare documents in a data frame for modeling with gibbs_sldax()
.post_regression()
to summarize regression relationships for objects of class Mlr
, Logistic
, or Sldax
: closes (#7).NEWS.md
file to track changes to the package.spelling
.knitr
v1.22 or higher.lda
testthat
v3.0.2 or higher.rmarkdown
.roxygen2
v7.1.2 with Markdown support for documentation.en-US
.Sldax
objects moved to a separate file.gg_coef()
; this function will be removed in a future release.stop()
and warning()
in the event of errors or warnings instead of print()
and cat()
.Rcpp_cout
and Rcpp_cerr
for printing messages and errors from compiled functions.get_zbar.Sldax()
was missing a return statement.OpenMP
as OpenMP
is not currently used.get_coherence()
computes the coherence score for each topic,get_exclusivity()
computes the exclusivity score for each topic.gibb_sldax()
gains the argument correct_ls
(TRUE
/ FALSE
),@extra$corrected_label_switching
for objects of class Model
.gibbs_sldax()
:sample_theta
(default = TRUE
) for the topic proportions,sample_beta
(default = TRUE
) for the topic-word probabilities.gibbs_sldax()
.return_assignments
(default = FALSE
).FALSE
is dramatically more memory efficient: closes (#37).docs
and data
arguments of gibbs_sldax()
are equal since missing data in either argument is currently not supported: closes (#36).my_fit
, use eta(my_fit)
.Model
: a generic model class,Mlr
: a linear regression class which inherits from Model
,Logistic
: a logistic regression class which inherits from Model
,Sldax
: a topic model class which inherits from Mlr
and Logistic
.label.switching
package.gibb_sldax()
for the Dirichlet priors on the topic proportions and on the topic-word probabilities to 1.0 (flat over the simplex).verbose = TRUE
specified to gibbs_sldax()
.gibbs_*()
functions where interrupting the process (e.g., hitting the Esc
key while running) and then trying to run these functions again would fail with the error ERROR: there is already an InterruptableProgressMonitor instance defined
.RcppProgress
v0.4.2.gibbs_logistic()
where regression coefficients could start with infinite values and the algorithm could not recover: closes (#39).gibbs_*()
functions.gibbs_sldax()
with constrain_eta = TRUE
and now warns user if exceeded.Rcpp
, RcppArmadillo
, and RcppProgress
packages.ccache
.get_zbar()
..est_betak()
, .est_thetad()
, .count_topic_word()
, .gibbs_mlr_cpp()
, .gibbs_logistic_cpp()
, .gibbs_sldax_cpp()
, rmvnorm_cpp()
, pwaic_d()
, waic_d()
.get_toptopics()
no longer requires a Sldax
model object and instead needs a topic-proportions matrix for all documents ('theta'): closes (#32).LDA
, Slda
, Sldalogit
) and reorganized Sldax
class.est_beta()
/est_theta()
: remove unwanted dim name in returned matrix; handle error in case where chain was shorter than 10 iterations.gg_coef()
no longer automatically prints a plot if the result of gg_coef()
is assigned to an object.term_score()
for 2-3x speed improvement, especially for large matrices.is()
from methods
package.Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.