From 88849be72a10ab5c6602a406f62df343c86c54ee Mon Sep 17 00:00:00 2001
From: Daniel <mail@danielluedecke.de>
Date: Fri, 29 Mar 2024 18:27:27 +0100
Subject: [PATCH] fix

---
 R/check_predictions.R | 31 +++++++++++++++++++++++++++++--
 1 file changed, 29 insertions(+), 2 deletions(-)

diff --git a/R/check_predictions.R b/R/check_predictions.R
index 82af994dd..6a7654096 100644
--- a/R/check_predictions.R
+++ b/R/check_predictions.R
@@ -161,10 +161,37 @@ check_predictions.stanreg <- function(object,
   )
 
   if (inherits(object, "brmsfit")) {
-    out <- bayesplot::pp_check(object, type = type, ndraws = iterations, ...)
+    out <- as.data.frame(bayesplot::pp_check(object, type = type, ndraws = iterations, ...))
   } else {
-    out <- bayesplot::pp_check(object, type = type, nreps = iterations, ...)
+    out <- as.data.frame(bayesplot::pp_check(object, type = type, nreps = iterations, ...))
   }
+
+  # bring data into shape, like we have for other models with `check_predictions()`
+  if (type == "density") {
+    d <- as.data.frame(out$data)
+    d_filter <- d[!d$is_y, ]
+    d_filter <- datawizard::data_to_wide(
+      d_filter,
+      id_cols = "y_id",
+      values_from = "value",
+      names_from = "rep_id"
+    )
+    d_filter$y_id <- NULL
+    colnames(d_filter) <- paste0("sim_", colnames(d_filter))
+    d_filter$y <- d$value[d$is_y, ]
+    out <- d_filter
+  } else {
+    colnames(out) <- c("x", "y", "CI_low", "Mean", "CI_high")
+  }
+
+  attr(out, "is_stan") <- TRUE
+  attr(out, "check_range") <- check_range
+  attr(out, "response_name") <- resp_string
+  attr(out, "bandwidth") <- bandwidth
+  attr(out, "model_info") <- minfo
+  attr(out, "type") <- type
+  class(out) <- c("performance_pp_check", "see_performance_pp_check", class(out))
+  out
 }
 
 #' @export