Deprecated functions and changes in selectors functions
all_categorical() are now compatible with gtsummary
version ≥ 2.0.0 (#270)scope_tidy() to scope a tidy tibble allowing
to tidy select (#270).select_to_varnames(), .formula_list_to_named_list(),
.generic_selector() and .is_selector_scoped() are now deprecated and
will be removed in a future release: you may consider
cards::process_selectors() and cards::process_formula_selectors() as
alternatives (#270)Fixes
model_get_model_frame.coxph() has been fixed to return a correct model
frame a subject identifier is passed to survival::coxph() (#268)Documentation
New features
model_matrix_attr in tidy_and_attach() and tidy_plus_plus()
to attach model frame and model matrix to the model as attributes for saving
some execution time (#254)tidy_add_n() now returns n_ind the number of individuals, in addition to
the number of observations (#251)tidy_parameters() calls now parameters::model_parameters()
with pretty_names = FALSE for saving execution time (#259)|>), requiring therefore R >= 4.2
(#262)Deprecated support
biglmm::bigglm() not supported anymore as biglmm has been removed from
CRANDeprecated functions
tidy_marginal_means() is now deprecated, following deprecation of
marginaleffects::marginal_means(). Use instead tidy_marginal_predictions()
with the option newdata = "marginalmeans".tidy_margins() is now indicated as superseded and may be deprecated if
margins is removed from CRAN. tidy_avg_slopes() could be used as an
alternative. (#252)Fixes
tidy_multgee() has been fixed to properly identify the different y.levels
(#260 @jackmwolf)tidy_marginal_predictions() has been updated to avoid the use of the
deprecated function marginaleffects::datagridcf() (#256)New supported models
New features
tidy_post_fun argument in tidy_plus_plus() (#235)Fix
tidy_marginal_predictions() (#245)supported_modelstidy_parameters() for mixed models (#238)survey::svyglm() models with replicate weights (#240)New features
MASS::contr.sdif() contrasts (#230)pscl::zeroinfl() and pscl::hurdle() models (#232)betareg::betareg() models (#234)Fix
packageVersion() should be a character string (#225)New features
tidy_add_estimate_to_reference_rows() now also populate p-values and
confidence intervals for sum contrasts (#220)nnet::multinom(), MASS::polr(),
ordinal::clm() and ordinal::clmm() models, as long as the type of models
is supported by the corresponding package, for example, margins does not
currently support nnet::multinom() models (#215)Improvements
marginaleffects version 0.10.0 (#216)New features
broom.helpers functions.tidy_marginal_predictions(),
plot_marginal_predictions(), tidy_all_effects(), and tidy_ggpredict()tidy_marginal_means()tidy_avg_comparisons() and
tidy_marginal_contrasts()tidy_avg_slopes() and tidy_margins()model_list_higher_order_variables() to list the highest order
combinations of variables (#202)model_get_response_variable() to get the name of the response
variable (#202)seq_range() to generate a sequence of values between
the minimum and the maximum of a vector (#202)contrasts_adjust in tidy_plus_plus(),
tidy_add_pairwise_contrasts() and model_get_pairwise_contrasts() allowing
to change the adjustment method used to compute pairwise contrasts (#204)New features
tidy_add_pairwise_contrasts() and model_get_pairwise_contrasts()
to compute pairwise contrasts of categorical variables with emmeans,
and corresponding new arguments in tidy_plus_plus() (#192)tidy_margins() to display Average Marginal Effects (#195)tidy_all_effects() to display Marginal Predictions (#195)tidy_ggpredict() to display Conditional Predictions (#195)Bug fixes and improvements
exponentiate argument is not appropriate (#197)New features
tidy_select_variables() now sorts the variables according to include (#183)New supported models
logitr::logitr() models (#179)multgee::nomLORgee() and multgee::ordLORgee()
models (#185)Bug fixes and improvements
.get_package_dependencies() to be more efficient. It now
looks only at a single package description file (#178).get_all_packages_dependencies() to list all dependencies of
all packages (#178).get_min_version_required() (#181)New features
.get_package_dependencies() listing all dependencies, including
minimum version required, of a package. (#171).assert_package() now taking into account the comparison
operator (> or >=) when a minimum version is required (#171)Bug fixes and improvements
tidyselect v1.2.0 (#173)mgcv::gam() models (#175)New supported models
parsnip::model_fit objects (#161)biglm::bigglm() and biglmm::bigglm() models (#155)fixest::feglm(), fixest::femlm(), fixest::feols()
and fixest::feNmlm() (requires R>=4.1) (#167)New features
dplyr::vars() (also exported by {gtsummary}) as a selector has
now been deprecated. Users will be warned that support for vars() will
eventually be removed from the package (#154).is_selector_scoped(), an internal function used in generating custom
selector functions, is now exported (#163)New features
.assert_package() now uses rlang::check_installed() and
rlang::is_installed() to check whether needed packages are installed. The
rlang::check_installed() prompts user to install needed package when run
interactively. (#147)tidy_add_n() and model_get_n() support for tidycmprsk::crr()
models (#143)supported_models tibble (#145)Bug fixes
tidy_add_n() to a mgcv::gam()
model with smooth terms (#150)New supported models
plm::plm() models (#140)New features
.formula_list_to_named_list() now respects the select_single=
argument for all inputs types. Previously, named lists were ignored..formula_list_to_named_list(null_allowed=) argument
that works in conjunction with type_check= asserting the class/type of
the RHS of the formula (or the value of the named list) (#137).formula_list_to_named_list() (#136)all_ran_pars() and all_ran_vals()Bug fixes
.formula_list_to_named_list(type_check=) checks RHS of a formula
and the value of named list. (#138)New features
model_get_coefficients_type.tidycrr() (#128)broom.helpers::tidy_parameters()
to include the package prefix. This message sometimes appears while
running gtsummary::tbl_regression() where some users may not be
aware where the tidy_paramters() function lives. (#129).formula_list_to_named_list() improvement: it is now possible to add
a type check (#132).assert_package() and .get_min_version_required() to
check for a package's installation status and whether the installed
version meets the minimum required version from the DESCRIPTION file (#134)Bug fixes
nnet::multinom() models with a binary outcome (#130)New supported models
glmmTMB::glmmTMB() models (#119)New features
list(trt ~ 1, all_continuous() ~ 2).~ <value> may be now used to indicate everything() ~ <value>Bug fixes
lme4::glmer() (#116)New supported models
rstanarm::stan_glm() modelsVGAM::vglm() models (#105)New features
tidy_parameters() based on parameters::model_parameters()
(#104)tidy_with_broom_or_parameters() (#104)tidy_plus_plus() now uses tidy_with_broom_or_parameters()model_get_coefficients_type() now returns "prop_hazard" for cloglog-binomial
models (#106)Bug fixes
tidy_add_n() works with multinomial models when y is not coded as
a factor (#109)glue added to SuggestsNew features
model_get_coefficients_type() now returns "relative_risk" for log-binomial
models (#101)tidy_disambiguate_terms() for disambiguating random-effect
terms in mixed models and new options for tidy_plus_plus():
disambiguate_terms (TRUE by default) and disambiguate_sep (#98)var_type column is now equal to "ran_pars" or
"ran_vals" for random-effect parameters and values, based of the
effect column returned by broom.mixed::tidy() (#90)model_list_contrasts()tidy_add_n() to add the number of observations (and for
relevant models the number of events and exposure time) (#64)add_n in tidy_plus_plus() (#64)model_get_n(), model_get_weights(), model_get_offset(),
model_get_response() and model_compute_terms_contributions() (#64)New supported models
lfe::felm() models (#79)brms::brm() models (#89)cmprsk::crr() models (#91)stats::nls() models (#97)mgcv::gam() models. (#82)Bug fixes and other changes
strict argument removed from
tidy_identify_variables() (#99)usethis::ui_*() messaging with cli::cli_*() (#94)tidy_add_term_labels() for variables with non standard
names (#77)rmarkdown (#95)var_type returned by
tidy_identify_variables() is now equal to "dichotomous" for categorical
variables with only 2 levelstidy_identify_variables()
now populates variable column by term content, instead of NA (#66)tidy_identify_variables(), the variable column is now populated
with the content of the term column (#63).generic_selector(): makes it easy to create selecting functions like
all_continuous()..select_to_varnames(): converts selecting syntax into character varnames.formula_list_to_named_list(): takes the formula selecting syntax and
converts it to a named list.all_continuous(), all_categorical(),
all_dichotomous(), all_contrasts(), all_intercepts() and
all_interaction() for selecting variables from models (#54)tidy_fun= (#49 @ddsjoberg)tidy_select_variables() to keep/drop
selected variables in the output (#45)tidy_add_coefficients_type() and
model_get_coefficients_type to get the type of coefficients
(generic, logistic, Poisson or proportional hazard) used
by a model (#46)tidy_add_contrasts() and model_list_contrasts() now return an
additional column contrasts_typeno_reference_row argument for tidy_add_reference_rows() (#47)model_get_nlevels to get the number of levels of categorical
variablesvar_nlevels returned by tidy_identify_variables(),
model_identify_variables() and tidy_plus_plus()model_list_terms_levels() and categorical_terms_pattern in
tidy_plus_plus() and tidy_add_term_labels (#61)model_list_terms_levels() now returns additional columns (level,
reference_level, contrasts_type, var_label, var_levels and
dichotomous)model_list_variables() now returns an additional var_label columnexponentiate argument is now passed to the tidy_*()
functions, as an attribute attached to the tibble, as well as custom labels
(variable_labels and term_labels)show_single_row argument now accepts tidyselect notation (#51 @ddsjoberg)tidy_add_estimate_to_reference_rows() now relies on emmeans for
sum contrasts, allowing to cover a wider range of modelstidy_* functions also inherits of "broom.helpers"
class (#56)interaction_sep argument has been added to tidy_plus_plus().clean_backticks() and .escape_regex() are now exportedtidy_add_header_rows() for nnet::multinom models:
label for header rows was missing (#50)tidy_identify_variables() correctly identify class "integer"
for this type of variables (#57)tidy_add_header_rows() for continuous variables with a non
standard name (#70)