Merge pull request #311 from stan-dev/prepare-v1.11.0

Prepare v1.11.0

DESCRIPTION

@@ -1,8 +1,8 @@
 Package: bayesplot
 Type: Package
 Title: Plotting for Bayesian Models
-Version: 1.10.0.9001
-Date: 2023-03-16
+Version: 1.11.0
+Date: 2024-01-30
 Authors@R: c(person("Jonah", "Gabry", role = c("aut", "cre"), email = "[email protected]"),
              person("Tristan", "Mahr", role = "aut"),
              person("Paul-Christian", "Bürkner", role = "ctb"),
@@ -29,8 +29,8 @@ Depends:
     R (>= 3.1.0)
 Imports:
     dplyr (>= 0.8.0),
-    ggplot2 (>= 3.0.0),
-    ggridges,
+    ggplot2 (>= 3.4.0),
+    ggridges (>= 0.5.5),
     glue,
     posterior,
     reshape2,

NAMESPACE

@@ -2,16 +2,31 @@
 
 S3method("[",neff_ratio)
 S3method("[",rhat)
+S3method(apply_transformations,array)
+S3method(apply_transformations,matrix)
+S3method(diagnostic_factor,neff_ratio)
+S3method(diagnostic_factor,rhat)
 S3method(log_posterior,CmdStanMCMC)
 S3method(log_posterior,stanfit)
 S3method(log_posterior,stanreg)
+S3method(melt_mcmc,matrix)
+S3method(melt_mcmc,mcmc_array)
 S3method(neff_ratio,CmdStanMCMC)
 S3method(neff_ratio,stanfit)
 S3method(neff_ratio,stanreg)
+S3method(num_chains,data.frame)
+S3method(num_chains,mcmc_array)
+S3method(num_iters,data.frame)
+S3method(num_iters,mcmc_array)
+S3method(num_params,data.frame)
+S3method(num_params,mcmc_array)
 S3method(nuts_params,CmdStanMCMC)
 S3method(nuts_params,list)
 S3method(nuts_params,stanfit)
 S3method(nuts_params,stanreg)
+S3method(parameter_names,array)
+S3method(parameter_names,default)
+S3method(parameter_names,matrix)
 S3method(plot,bayesplot_grid)
 S3method(plot,bayesplot_scheme)
 S3method(pp_check,default)

NEWS.md

@@ -1,9 +1,16 @@
 <!-- See http://style.tidyverse.org/news.html for advice on writing news -->
 
-# bayesplot 1.10.0.9000
-
-* `ppc_pit_ecdf()` and `ppc_pit_ecdf_grouped()` now support discrete variables, and their default method for selecting the number of ECDF evaluation points has been updated.
-* Items for next release here
+# bayesplot 1.11.0
+
+* Update for new ggplot2 release by @teunbrand in #309
+* Add `bins` argument to many histogram plots by @paul-buerkner in #300
+* Follow ggplot2 updates on `facet_grid()` and `facet_wrap()` by @heavywatal in #305
+* Better `ppc_loo_pit_qq` plots by @avehtari in #307
+* Check `prob` is numeric for intervals plots by @tony-stone in #299
+* Add `bins` and `breaks` arguments to more histogram and hex plots by @heavywatal in #313
+* Replace `size` argument with `linewidth` for `geom_line` and `geom_ridgeline` by @heavywatal in #314
+* All LOO plots now accept `psis_object` argument by @jgabry in #311
+* `ppc_pit_ecdf()` and `ppc_pit_ecdf_grouped()` now support discrete variables, and their default method for selecting the number of ECDF evaluation points has been updated. by @TeemuSailynoja in #316
 
 # bayesplot 1.10.0
 
@@ -49,14 +56,14 @@ previous releases, but the new ones in this release are:
    - `ppc_error_hist_grouped()`
    - `ppc_error_scatter()`
    - `ppc_error_binned()`
-   
+
 * New plotting function `ppc_km_overlay_grouped()`, the grouped variant of
   `ppc_km_overlay()`. (#260, @fweber144)
 
 * `ppc_scatter()`, `ppc_scatter_avg()`, and `ppc_scatter_avg_grouped()` gain an
   argument `ref_line`, which can be set to `FALSE` to turn off the `x=y` line
   drawn behind the scatterplot.
-  
+
 * `ppc_ribbon()` and `ppc_ribbon_grouped()` gain argument `y_draw` that specifies whether the observed y should be plotted using a point, line, or both. (#257, @charlesm93)
 
 * `mcmc_*()` functions now support all draws formats from the **posterior** package. (#277, @Ozan147)

R/helpers-mcmc.R

@@ -51,13 +51,13 @@ prepare_mcmc_array <- function(x,
   if (is.matrix(x)) {
     x <- x[, pars, drop=FALSE]
     if (length(transformations)) {
-      x <- apply_transformations(x, transformations)
+      x <- apply_transformations(x, transformations = transformations)
     }
     x <- array(x, dim = c(nrow(x), 1, ncol(x)))
   } else {
     x <- x[, , pars, drop = FALSE]
     if (length(transformations)) {
-      x <- apply_transformations(x, transformations)
+      x <- apply_transformations(x, transformations = transformations)
     }
   }
 
@@ -124,6 +124,8 @@ select_parameters <-
 #' @return A molten data frame.
 #'
 melt_mcmc <- function(x, ...) UseMethod("melt_mcmc")
+
+#' @export
 melt_mcmc.mcmc_array <- function(x,
                                  varnames =
                                    c("Iteration", "Chain", "Parameter"),
@@ -144,6 +146,7 @@ melt_mcmc.mcmc_array <- function(x,
 }
 
 # If all chains are already merged
+#' @export
 melt_mcmc.matrix <- function(x,
                              varnames = c("Draw", "Parameter"),
                              value.name = "Value",
@@ -305,13 +308,17 @@ chain_list2array <- function(x) {
 
 # Get parameter names from a 3-D array
 parameter_names <- function(x) UseMethod("parameter_names")
+
+#' @export
 parameter_names.array <- function(x) {
   stopifnot(is_3d_array(x))
   dimnames(x)[[3]] %||% abort("No parameter names found.")
 }
+#' @export
 parameter_names.default <- function(x) {
   colnames(x) %||% abort("No parameter names found.")
 }
+#' @export
 parameter_names.matrix <- function(x) {
   colnames(x) %||% abort("No parameter names found.")
 }
@@ -388,10 +395,12 @@ validate_transformations <-
 #'   functions.
 #' @return x, with tranformations having been applied to some parameters.
 #'
-apply_transformations <- function(x, transformations = list(), ...) {
+apply_transformations <- function(x, ...) {
   UseMethod("apply_transformations")
 }
-apply_transformations.matrix <- function(x, transformations = list()) {
+
+#' @export
+apply_transformations.matrix <- function(x, ..., transformations = list()) {
   pars <- colnames(x)
   x_transforms <- validate_transformations(transformations, pars)
   for (p in names(x_transforms)) {
@@ -400,7 +409,9 @@ apply_transformations.matrix <- function(x, transformations = list()) {
 
   x
 }
-apply_transformations.array <- function(x, transformations = list()) {
+
+#' @export
+apply_transformations.array <- function(x, ..., transformations = list()) {
   stopifnot(length(dim(x)) == 3)
   pars <- dimnames(x)[[3]]
   x_transforms <- validate_transformations(transformations, pars)
@@ -437,17 +448,23 @@ num_chains <- function(x, ...) UseMethod("num_chains")
 num_iters <- function(x, ...) UseMethod("num_iters")
 num_params <- function(x, ...) UseMethod("num_params")
 
+#' @export
 num_params.mcmc_array <- function(x, ...) dim(x)[3]
+#' @export
 num_chains.mcmc_array <- function(x, ...) dim(x)[2]
+#' @export
 num_iters.mcmc_array <- function(x, ...) dim(x)[1]
+#' @export
 num_params.data.frame <- function(x, ...) {
   stopifnot("Parameter" %in% colnames(x))
   length(unique(x$Parameter))
 }
+#' @export
 num_chains.data.frame <- function(x, ...) {
   stopifnot("Chain" %in% colnames(x))
   length(unique(x$Chain))
 }
+#' @export
 num_iters.data.frame <- function(x, ...) {
   cols <- colnames(x)
   stopifnot("Iteration" %in% cols || "Draws" %in% cols)

R/mcmc-diagnostics.R

@@ -360,17 +360,19 @@ mcmc_acf_bar <-
 #'   `x <= breaks[1]`, `breaks[1] < x <= breaks[2]`, `x > breaks[2]`).
 #' @return A factor the same length as `x` with three levels.
 #' @noRd
-diagnostic_factor <- function(x, breaks, ...) {
+diagnostic_factor <- function(x, ...) {
   UseMethod("diagnostic_factor")
 }
 
-diagnostic_factor.rhat <- function(x, breaks = c(1.05, 1.1)) {
+#' @export
+diagnostic_factor.rhat <- function(x, ..., breaks = c(1.05, 1.1)) {
   cut(x, breaks = c(-Inf, breaks, Inf),
       labels = c("low", "ok", "high"),
       ordered_result = FALSE)
 }
 
-diagnostic_factor.neff_ratio <- function(x, breaks = c(0.1, 0.5)) {
+#' @export
+diagnostic_factor.neff_ratio <- function(x, ..., breaks = c(0.1, 0.5)) {
   cut(x, breaks = c(-Inf, breaks, Inf),
       labels = c("low", "ok", "high"),
       ordered_result = FALSE)

R/ppc-discrete.R

@@ -367,7 +367,7 @@ ppc_bars_data <-
 #' @param y,yrep,group User's already validated `y`, `yrep`, and (if applicable)
 #'   `group` arguments.
 #' @param prob,freq User's `prob` and `freq` arguments.
-#' @importFrom dplyr "%>%" ungroup count arrange mutate summarise across full_join rename all_of
+#' @importFrom dplyr %>% ungroup count arrange mutate summarise across full_join rename all_of
 .ppc_bars_data <- function(y, yrep, group = NULL, prob = 0.9, freq = TRUE) {
   alpha <- (1 - prob) / 2
   probs <- sort(c(alpha, 0.5, 1 - alpha))