Factor rvars

.github/workflows/rcmdcheck.yml

@@ -23,10 +23,10 @@ jobs:
           - {os: macOS-latest,   r: 'devel', suggested_check: TRUE}
           - {os: macOS-latest,   r: 'release', suggested_check: TRUE}
           - {os: windows-latest, r: 'release', suggested_check: TRUE}
-          - {os: windows-latest, r: '3.6', suggested_check: FALSE}
           - {os: ubuntu-18.04,   r: 'devel',   rspm: "https://packagemanager.rstudio.com/cran/__linux__/bionic/latest", http-user-agent: "R/4.0.0 (ubuntu-18.04) R (4.0.0 x86_64-pc-linux-gnu x86_64 linux-gnu) on GitHub Actions" , suggested_check: TRUE}
           - {os: ubuntu-18.04,   r: 'release', rspm: "https://packagemanager.rstudio.com/cran/__linux__/bionic/latest", suggested_check: TRUE}
           - {os: ubuntu-18.04,   r: 'oldrel',  rspm: "https://packagemanager.rstudio.com/cran/__linux__/bionic/latest", suggested_check: FALSE}
+          - {os: ubuntu-18.04,   r: '3.6',  rspm: "https://packagemanager.rstudio.com/cran/__linux__/bionic/latest", suggested_check: FALSE}
 
     env:
       R_REMOTES_NO_ERRORS_FROM_WARNINGS: true

DESCRIPTION

@@ -31,7 +31,7 @@ Imports:
   rlang (>= 0.4.7),
   stats,
   tibble (>= 3.0.0),
-  vctrs,
+  vctrs (>= 0.5.0),
   tensorA,
   pillar,
   distributional,
@@ -53,5 +53,5 @@ LazyData: false
 URL: https://mc-stan.org/posterior/, https://discourse.mc-stan.org/
 BugReports: https://github.com/stan-dev/posterior/issues
 Roxygen: list(markdown = TRUE)
-RoxygenNote: 7.2.1
+RoxygenNote: 7.2.2
 VignetteBuilder: knitr

NAMESPACE

@@ -11,6 +11,7 @@ S3method("[[",rvar)
 S3method("[[<-",rvar)
 S3method("dim<-",rvar)
 S3method("dimnames<-",rvar)
+S3method("levels<-",rvar)
 S3method("names<-",rvar)
 S3method("variables<-",draws_array)
 S3method("variables<-",draws_df)
@@ -23,7 +24,10 @@ S3method(.subset_draws,draws_list)
 S3method(.subset_draws,draws_matrix)
 S3method(.subset_draws,draws_rvars)
 S3method(Math,rvar)
+S3method(Math,rvar_factor)
 S3method(Ops,rvar)
+S3method(Ops,rvar_factor)
+S3method(Ops,rvar_ordered)
 S3method(Pr,default)
 S3method(Pr,logical)
 S3method(Pr,rvar)
@@ -32,6 +36,7 @@ S3method(all,equal.rvar)
 S3method(all,rvar)
 S3method(any,rvar)
 S3method(anyDuplicated,rvar)
+S3method(anyDuplicated,rvar_factor)
 S3method(anyNA,rvar)
 S3method(aperm,rvar)
 S3method(as.data.frame,rvar)
@@ -91,9 +96,13 @@ S3method(bind_draws,draws_list)
 S3method(bind_draws,draws_matrix)
 S3method(bind_draws,draws_rvars)
 S3method(bind_draws,list)
+S3method(broadcast_and_bind_rvars,rvar)
+S3method(broadcast_and_bind_rvars,rvar_factor)
 S3method(c,rvar)
 S3method(cbind,rvar)
 S3method(cdf,rvar)
+S3method(cdf,rvar_factor)
+S3method(cdf,rvar_ordered)
 S3method(chain_ids,"NULL")
 S3method(chain_ids,draws_array)
 S3method(chain_ids,draws_df)
@@ -102,8 +111,11 @@ S3method(chain_ids,draws_matrix)
 S3method(chain_ids,draws_rvars)
 S3method(chol,rvar)
 S3method(density,rvar)
+S3method(density,rvar_factor)
 S3method(dim,rvar)
 S3method(dimnames,rvar)
+S3method(dissent,default)
+S3method(dissent,rvar)
 S3method(draw_ids,"NULL")
 S3method(draw_ids,draws_array)
 S3method(draw_ids,draws_df)
@@ -112,6 +124,9 @@ S3method(draw_ids,draws_matrix)
 S3method(draw_ids,draws_rvars)
 S3method(draw_ids,rvar)
 S3method(duplicated,rvar)
+S3method(duplicated,rvar_factor)
+S3method(entropy,default)
+S3method(entropy,rvar)
 S3method(ess_basic,default)
 S3method(ess_basic,rvar)
 S3method(ess_bulk,default)
@@ -132,6 +147,9 @@ S3method(extract_variable_matrix,draws)
 S3method(extract_variable_matrix,draws_rvars)
 S3method(format,rvar)
 S3method(format_glimpse,rvar)
+S3method(get_rvar_class,default)
+S3method(get_rvar_class,factor)
+S3method(get_rvar_class,ordered)
 S3method(is.array,rvar)
 S3method(is.finite,rvar)
 S3method(is.infinite,rvar)
@@ -149,6 +167,9 @@ S3method(length,rvar)
 S3method(levels,rvar)
 S3method(mad,default)
 S3method(mad,rvar)
+S3method(mad,rvar_ordered)
+S3method(match,default)
+S3method(match,rvar)
 S3method(max,rvar)
 S3method(mcse_mean,default)
 S3method(mcse_mean,rvar)
@@ -165,6 +186,8 @@ S3method(merge_chains,draws_matrix)
 S3method(merge_chains,draws_rvars)
 S3method(merge_chains,rvar)
 S3method(min,rvar)
+S3method(modal_category,default)
+S3method(modal_category,rvar)
 S3method(mutate_variables,draws_array)
 S3method(mutate_variables,draws_df)
 S3method(mutate_variables,draws_list)
@@ -209,6 +232,8 @@ S3method(print,draws_rvars)
 S3method(print,rvar)
 S3method(prod,rvar)
 S3method(quantile,rvar)
+S3method(quantile,rvar_factor)
+S3method(quantile,rvar_ordered)
 S3method(quantile2,default)
 S3method(quantile2,rvar)
 S3method(range,rvar)
@@ -262,6 +287,7 @@ S3method(t,rvar)
 S3method(thin_draws,draws)
 S3method(thin_draws,rvar)
 S3method(unique,rvar)
+S3method(unique,rvar_factor)
 S3method(var,default)
 S3method(var,rvar)
 S3method(variables,"NULL")
@@ -274,33 +300,79 @@ S3method(variance,draws_array)
 S3method(variance,draws_matrix)
 S3method(variance,rvar)
 S3method(vec_cast,character.rvar)
+S3method(vec_cast,character.rvar_factor)
+S3method(vec_cast,character.rvar_ordered)
 S3method(vec_cast,distribution.rvar)
+S3method(vec_cast,rvar.character)
 S3method(vec_cast,rvar.distribution)
 S3method(vec_cast,rvar.double)
+S3method(vec_cast,rvar.factor)
 S3method(vec_cast,rvar.integer)
 S3method(vec_cast,rvar.logical)
+S3method(vec_cast,rvar.ordered)
 S3method(vec_cast,rvar.rvar)
+S3method(vec_cast,rvar.rvar_factor)
+S3method(vec_cast,rvar.rvar_ordered)
+S3method(vec_cast,rvar_factor.character)
+S3method(vec_cast,rvar_factor.double)
+S3method(vec_cast,rvar_factor.factor)
+S3method(vec_cast,rvar_factor.integer)
+S3method(vec_cast,rvar_factor.logical)
+S3method(vec_cast,rvar_factor.ordered)
+S3method(vec_cast,rvar_factor.rvar)
+S3method(vec_cast,rvar_factor.rvar_factor)
+S3method(vec_cast,rvar_factor.rvar_ordered)
+S3method(vec_cast,rvar_ordered.character)
+S3method(vec_cast,rvar_ordered.double)
+S3method(vec_cast,rvar_ordered.factor)
+S3method(vec_cast,rvar_ordered.integer)
+S3method(vec_cast,rvar_ordered.logical)
+S3method(vec_cast,rvar_ordered.ordered)
+S3method(vec_cast,rvar_ordered.rvar)
+S3method(vec_cast,rvar_ordered.rvar_factor)
+S3method(vec_cast,rvar_ordered.rvar_ordered)
 S3method(vec_proxy,rvar)
 S3method(vec_ptype,rvar)
+S3method(vec_ptype,rvar_factor)
+S3method(vec_ptype,rvar_ordered)
+S3method(vec_ptype2,character.rvar_factor)
+S3method(vec_ptype2,character.rvar_ordered)
 S3method(vec_ptype2,distribution.rvar)
 S3method(vec_ptype2,double.rvar)
+S3method(vec_ptype2,factor.rvar_factor)
+S3method(vec_ptype2,factor.rvar_ordered)
 S3method(vec_ptype2,integer.rvar)
 S3method(vec_ptype2,logical.rvar)
+S3method(vec_ptype2,ordered.rvar_factor)
+S3method(vec_ptype2,ordered.rvar_ordered)
 S3method(vec_ptype2,rvar.distribution)
 S3method(vec_ptype2,rvar.double)
 S3method(vec_ptype2,rvar.integer)
 S3method(vec_ptype2,rvar.logical)
 S3method(vec_ptype2,rvar.rvar)
+S3method(vec_ptype2,rvar_factor.character)
+S3method(vec_ptype2,rvar_factor.factor)
+S3method(vec_ptype2,rvar_factor.ordered)
+S3method(vec_ptype2,rvar_factor.rvar_factor)
+S3method(vec_ptype2,rvar_ordered.character)
+S3method(vec_ptype2,rvar_ordered.factor)
+S3method(vec_ptype2,rvar_ordered.ordered)
+S3method(vec_ptype2,rvar_ordered.rvar_ordered)
 S3method(vec_ptype_abbr,rvar)
+S3method(vec_ptype_abbr,rvar_factor)
+S3method(vec_ptype_abbr,rvar_ordered)
 S3method(vec_ptype_full,rvar)
 S3method(vec_restore,rvar)
+S3method(vec_restore,rvar_factor)
+S3method(vec_restore,rvar_ordered)
 S3method(weight_draws,draws_array)
 S3method(weight_draws,draws_df)
 S3method(weight_draws,draws_list)
 S3method(weight_draws,draws_matrix)
 S3method(weight_draws,draws_rvars)
 S3method(weights,draws)
 export("%**%")
+export("%in%")
 export("draws_of<-")
 export("variables<-")
 export(E)
@@ -312,6 +384,9 @@ export(as_draws_list)
 export(as_draws_matrix)
 export(as_draws_rvars)
 export(as_rvar)
+export(as_rvar_factor)
+export(as_rvar_numeric)
+export(as_rvar_ordered)
 export(autocorrelation)
 export(autocovariance)
 export(bind_draws)
@@ -321,6 +396,7 @@ export(default_convergence_measures)
 export(default_mcse_measures)
 export(default_summary_measures)
 export(diag)
+export(dissent)
 export(draw_ids)
 export(draws_array)
 export(draws_df)
@@ -329,6 +405,7 @@ export(draws_matrix)
 export(draws_of)
 export(draws_rvars)
 export(drop)
+export(entropy)
 export(ess_basic)
 export(ess_bulk)
 export(ess_mean)
@@ -347,13 +424,17 @@ export(is_draws_list)
 export(is_draws_matrix)
 export(is_draws_rvars)
 export(is_rvar)
+export(is_rvar_factor)
+export(is_rvar_ordered)
 export(iteration_ids)
 export(mad)
+export(match)
 export(mcse_mean)
 export(mcse_median)
 export(mcse_quantile)
 export(mcse_sd)
 export(merge_chains)
+export(modal_category)
 export(mutate_variables)
 export(nchains)
 export(ndraws)
@@ -375,6 +456,7 @@ export(rvar)
 export(rvar_all)
 export(rvar_any)
 export(rvar_apply)
+export(rvar_factor)
 export(rvar_is_finite)
 export(rvar_is_infinite)
 export(rvar_is_na)
@@ -384,6 +466,7 @@ export(rvar_max)
 export(rvar_mean)
 export(rvar_median)
 export(rvar_min)
+export(rvar_ordered)
 export(rvar_prod)
 export(rvar_quantile)
 export(rvar_range)
@@ -438,7 +521,6 @@ importFrom(utils,lsf.str)
 importFrom(utils,str)
 importFrom(vctrs,new_vctr)
 importFrom(vctrs,vec_cast)
-importFrom(vctrs,vec_chop)
 importFrom(vctrs,vec_proxy)
 importFrom(vctrs,vec_ptype)
 importFrom(vctrs,vec_ptype2)

NEWS.md

@@ -2,8 +2,26 @@
 
 ### Enhancements
 
+* Added new `rvar_factor()` and `rvar_ordered()` subtypes of `rvar()` that work
+  analogously to `factor()` and `ordered()` (#149). See the new section on
+  `rvar_factor`s in `vignette("rvar")`.
+* The `draws_df()`, `draws_list()`, and `draws_rvars()` formats now support
+  discrete variables stored as `factors` / `ordered`s (or `rvar_factor`s /
+  `rvar_ordered`s). If converted to formats that do not support discrete
+  variables with named levels (`draws_matrix()` and `draws_array()`), 
+  factor-like variables are converted to `numeric`s.
+* Made `match()` and `%in%` generic and added support for `rvar`s to both
+  functions.
+* Added `modal_category()`, `entropy()`, and `dissent()` functions for
+  summarizing discrete draws.
 * Allow lists of draws objects to be passed as the first argument to 
   `bind_draws()` (#253).
+* `print.rvar()` and `format.rvar()` now default to a smaller number of
+  significant digits in more cases, including when printing in data frames.
+  This is controlled by the new `"posterior.digits"` option (see 
+  `help("posterior-package")`).
+* Implemented faster `vec_proxy.rvar()` and `vec_restore.rvar()`, improving
+  performance of `rvar`s in `tibble`s (and elsewhere `vctrs` is used).
 
 
 # posterior 1.3.1

R/as_draws_array.R

@@ -59,6 +59,11 @@ as_draws_array.draws_df <- function(x, ...) {
   iterations <- iteration_ids(x)
   chains <- chain_ids(x)
   out <- vector("list", length(chains))
+  non_numeric_cols <- vapply(x, function(x_i) !is.numeric(x_i) && !is.logical(x_i), logical(1))
+  if (any(non_numeric_cols)) {
+    warning_no_call("draws_array does not support non-numeric variables (e.g., factors). Converting non-numeric variables to numeric.")
+  }
+  x[, non_numeric_cols] <- lapply(unclass(x)[non_numeric_cols], as.numeric)
   for (i in seq_along(out)) {
     if (length(chains) == 1) {
       out[[i]] <- x
@@ -92,6 +97,11 @@ as_draws_array.draws_rvars <- function(x, ...) {
     return(empty_draws_array(variables(x)))
   }
 
+  factor_variables <- vapply(x, is_rvar_factor, logical(1))
+  if (any(factor_variables)) {
+    warning_no_call("draws_array does not support non-numeric variables (e.g., factors). Converting non-numeric variables to numeric.")
+  }
+
   draws <- do.call(cbind, lapply(seq_along(x), function(i) {
     # flatten each rvar so it only has two dimensions: draws and variables
     # this also collapses indices into variable names in the format "var[i,j,k,...]"

R/as_draws_df.R

@@ -95,8 +95,7 @@ as_draws_df.draws_list <- function(x, ...) {
   iteration_ids <- iteration_ids(x)
   chain_ids <- chain_ids(x)
   vars <- names(x[[1L]])
-  x <- do.call(rbind.data.frame, x)
-  colnames(x) <- vars
+  x <- do.call(vctrs::vec_rbind, lapply(x, vctrs::new_data_frame))
   x <- tibble::as_tibble(x)
   x[[".chain"]] <- rep(chain_ids, each = max(iteration_ids))
   x[[".iteration"]] <- rep(iteration_ids, max(chain_ids))
@@ -108,7 +107,21 @@ as_draws_df.draws_list <- function(x, ...) {
 #' @rdname draws_df
 #' @export
 as_draws_df.draws_rvars <- function(x, ...) {
-  as_draws_df(as_draws_array(x), ...)
+  if (ndraws(x) == 0L) {
+    return(empty_draws_df(variables(x)))
+  }
+  out <- do.call(cbind, lapply(seq_along(x), function(i) {
+    # flatten each rvar so it only has two dimensions: draws and variables
+    # this also collapses indices into variable names in the format "var[i,j,k,...]"
+    flatten_rvar_draws_to_df(x[[i]], names(x)[[i]])
+  }))
+  iteration_ids <- iteration_ids(x)
+  chain_ids <- chain_ids(x)
+  out[[".chain"]] <- rep(chain_ids, each = max(iteration_ids))
+  out[[".iteration"]] <- rep(iteration_ids, max(chain_ids))
+  out[[".draw"]] <- draw_ids(x)
+  class(out) <- class_draws_df()
+  out
 }
 
 #' @rdname draws_df

R/as_draws_list.R

@@ -68,7 +68,7 @@ as_draws_list.draws_df <- function(x, ...) {
 #' @rdname draws_list
 #' @export
 as_draws_list.draws_rvars <- function(x, ...) {
-  as_draws_list(as_draws_array(x), ...)
+  as_draws_list(as_draws_df(x), ...)
 }
 
 #' @rdname draws_list