Prepare CRAN release (#1106)

* Test on latest marginaleffects

* fix

* fix

* use latest CRAN versions

* Update .Rbuildignore

* docs

* add option to include intercept

* Test with effectsize dev

* remove remotes

Author: strengejacke

.Rbuildignore

@@ -20,6 +20,7 @@
 ^vignettes/p.
 ^vignettes/s.
 ^tests/_snaps/.
+^tests/testthat/_snaps/.
 ^cran-comments\.md$
 ^revdep$
 ^CODE_OF_CONDUCT\.md$

DESCRIPTION

@@ -1,7 +1,7 @@
 Type: Package
 Package: parameters
 Title: Processing of Model Parameters
-Version: 0.25.0.6
+Version: 0.26.0
 Authors@R:
     c(person(given = "Daniel",
              family = "Lüdecke",
@@ -78,9 +78,9 @@ BugReports: https://github.com/easystats/parameters/issues
 Depends:
     R (>= 3.6)
 Imports:
-    bayestestR (>= 0.15.3),
+    bayestestR (>= 0.16.0),
     datawizard (>= 1.0.2),
-    insight (>= 1.2.0),
+    insight (>= 1.3.0),
     graphics,
     methods,
     stats,
@@ -156,7 +156,7 @@ Suggests:
     logspline,
     lqmm,
     M3C,
-    marginaleffects (>= 0.25.0),
+    marginaleffects (>= 0.26.0),
     modelbased (>= 0.9.0),
     MASS,
     Matrix,
@@ -227,4 +227,3 @@ Config/testthat/parallel: true
 Config/Needs/website: easystats/easystatstemplate
 Config/Needs/check: stan-dev/cmdstanr
 Config/rcmdcheck/ignore-inconsequential-notes: true
-Remotes: easystats/insight, easystats/bayestestR

NEWS.md

@@ -1,4 +1,4 @@
-# parameters (devel)
+# parameters 0.26.0
 
 ## Changes
 
@@ -11,6 +11,18 @@
   p-adjustment was applied during model-fitting, the correct p-values are now
   returned (before, unadjusted p-values were returned in some cases).
 
+* Revised code-base to address changes in latest *insight* update. Dealing with
+  larger models (many parameters, many posterior samples) from packages *brms*
+  and *rstanarm* is more efficient now. Furthermore, the options for the
+  `effects` argument have a new behaviour. `"all"` only returns fixed effects
+  and random effects variance components, but no longer the group level
+  estimates. Use `effects = "full"` to return all parameters. This change is
+  mainly to be more flexible and gain more efficiency for models with many
+  parameters and / or many posterior draws.
+
+* `model_parameters()` for Anova objects gains an `include_intercept` argument,
+  to include intercepts in the Anova table, where possible.
+
 # parameters 0.25.0
 
 ## Changes

R/extract_parameters_anova.R

@@ -2,6 +2,7 @@
 .extract_parameters_anova <- function(model,
                                       test = "multivariate",
                                       p_adjust = NULL,
+                                      include_intercept = FALSE,
                                       verbose = TRUE) {
   # Processing
   if (inherits(model, "manova")) {
@@ -24,7 +25,7 @@
 
   # remove intercept
   intercepts <- parameters$Parameter %in% c("Intercept", "(Intercept)")
-  if (any(intercepts)) {
+  if (any(intercepts) && !include_intercept) {
     parameters <- parameters[!intercepts, ]
   }
 

R/methods_aov.R

@@ -4,9 +4,8 @@
 
 #' Parameters from ANOVAs
 #'
-#' @param model Object of class [aov()], [anova()],
-#'   `aovlist`, `Gam`, [manova()], `Anova.mlm`,
-#'   `afex_aov` or `maov`.
+#' @param model Object of class [aov()], [anova()], `aovlist`, `Gam`,
+#'   [manova()], `Anova.mlm`, `afex_aov` or `maov`.
 #' @param es_type The effect size of interest. Not that possibly not all
 #'   effect sizes are applicable to the model object. See 'Details'. For Anova
 #'   models, can also be a character vector with multiple effect size names.
@@ -29,6 +28,8 @@
 #' @param table_wide Logical that decides whether the ANOVA table should be in
 #'   wide format, i.e. should the numerator and denominator degrees of freedom
 #'   be in the same row. Default: `FALSE`.
+#' @param include_intercept Logical, if `TRUE`, includes the intercept
+#'   (`(Intercept)`) in the anova table.
 #' @param alternative A character string specifying the alternative hypothesis;
 #'   Controls the type of CI returned: `"two.sided"` (default, two-sided CI),
 #'   `"greater"` or `"less"` (one-sided CI). Partial matching is allowed
@@ -108,6 +109,7 @@ model_parameters.aov <- function(model,
                                  es_type = NULL,
                                  keep = NULL,
                                  drop = NULL,
+                                 include_intercept = FALSE,
                                  table_wide = FALSE,
                                  verbose = TRUE,
                                  ...) {
@@ -139,7 +141,13 @@ model_parameters.aov <- function(model,
   }
 
   # extract standard parameters
-  params <- .extract_parameters_anova(model, test, p_adjust = p_adjust, verbose = verbose)
+  params <- .extract_parameters_anova(
+    model,
+    test,
+    p_adjust = p_adjust,
+    include_intercept = include_intercept,
+    verbose = verbose
+  )
 
   # add effect sizes, if available
   params <- .effectsizes_for_aov(
@@ -252,6 +260,7 @@ model_parameters.afex_aov <- function(model,
                                       type = NULL,
                                       keep = NULL,
                                       drop = NULL,
+                                      include_intercept = FALSE,
                                       p_adjust = NULL,
                                       verbose = TRUE,
                                       ...) {
@@ -260,11 +269,23 @@ model_parameters.afex_aov <- function(model,
     with_df_and_p <- summary(model$Anova)$univariate.tests
     params$`Sum Sq` <- with_df_and_p[-1, 1]
     params$`Error SS` <- with_df_and_p[-1, 3]
-    out <- .extract_parameters_anova(params, test = NULL, p_adjust = NULL, verbose)
+    out <- .extract_parameters_anova(
+      params,
+      test = NULL,
+      include_intercept = include_intercept,
+      p_adjust = NULL,
+      verbose
+    )
     p_adjust <- .extract_p_adjust_afex(model, p_adjust)
   } else {
     p_adjust <- .extract_p_adjust_afex(model, p_adjust)
-    out <- .extract_parameters_anova(model$Anova, test = NULL, p_adjust, verbose)
+    out <- .extract_parameters_anova(
+      model$Anova,
+      test = NULL,
+      include_intercept = include_intercept,
+      p_adjust,
+      verbose
+    )
   }
 
   out <- .effectsizes_for_aov(

R/methods_brms.R

@@ -12,9 +12,11 @@
 #' @param ci Credible Interval (CI) level. Default to `0.95` (`95%`). See
 #' [bayestestR::ci()] for further details.
 #' @param group_level Logical, for multilevel models (i.e. models with random
-#' effects) and when `effects = "all"` or `effects = "random"`, include the
-#' parameters for each group level from random effects. If `group_level = FALSE`
-#' (the default), only information on SD and COR are shown.
+#' effects) and when `effects = "random"`, return the parameters for each group
+#' level from random effects only. If `group_level = FALSE` (the default), also
+#' information on SD and COR are returned. Note that this argument is superseded
+#' by the new options for the `effects` argument. `effects = "grouplevel"` should
+#' be used instead of `group_level = TRUE`.
 #' @param component Which type of parameters to return, such as parameters for the
 #' conditional model, the zero-inflation part of the model, the dispersion
 #' term, or other auxiliary parameters be returned? Applies to models with

R/methods_glmmTMB.R

@@ -40,6 +40,10 @@
 #'   computed by default, for simpler models or fewer observations, confidence
 #'   intervals will be included. Set explicitly to `TRUE` or `FALSE` to enforce
 #'   or omit calculation of confidence intervals.
+#' @param group_level Logical, for multilevel models (i.e. models with random
+#'   effects) and when `effects = "random"`, include the parameters for each
+#'   group level from random effects. If `group_level = FALSE` (the default),
+#'   only information on SD and COR are shown.
 #' @param ... Arguments passed to or from other methods. For instance, when
 #'   `bootstrap = TRUE`, arguments like `type` or `parallel` are passed down to
 #'   `bootstrap_model()`.