merge pr #89: accommodate outcome levels with special characters

NEWS.md

@@ -2,6 +2,9 @@
 
 To released as v0.2.2.
 
+* Fixed bug arising from outcome levels that are not valid column 
+  names in the multinomial classification setting. 
+
 # v0.2.1
 
 * Updates for importing workflow sets that use the `add_variables()` 

R/add_candidates.R

@@ -281,7 +281,8 @@ add_candidates.default <- function(data_stack, candidates, name, ...) {
 
   pred_class_idx <- grepl(pattern = ".pred_class", x = colnames(candidate_cols))
 
-  candidate_cols <- candidate_cols[,!pred_class_idx]
+  candidate_cols <- candidate_cols[,!pred_class_idx] %>% 
+    setNames(., make.names(names(.)))
 
   if (nrow(stack) == 0) {
     stack <- 
@@ -473,7 +474,9 @@ process_.config <- function(.config, df, name) {
 
 # For racing, we only want to keep the candidates with complete resamples. 
 collate_predictions <- function(x) {
-  res <- tune::collect_predictions(x, summarize = TRUE)
+  res <- tune::collect_predictions(x, summarize = TRUE) %>%
+    dplyr::rename_with(make.names, .cols = dplyr::starts_with(".pred"))
+
   if (inherits(x, "tune_race")) {
     config_counts <- 
       tune::collect_metrics(x, summarize = FALSE) %>% 

R/fit_members.R

@@ -181,7 +181,8 @@ sanitize_classification_names <- function(model_stack, member_names) {
     as.character() %>%
     unique()
 
-  pred_strings <- paste0(".pred_", outcome_levels, "_")
+  pred_strings <- paste0(".pred_", outcome_levels, "_") %>% 
+    make.names()
 
   new_member_names <-
     gsub(

R/predict.R

@@ -93,7 +93,8 @@ predict.model_stack <- function(object, new_data, type = NULL, members = FALSE,
       opts = opts,
       type = member_type
     ) %>%
-    rlang::eval_tidy()
+    rlang::eval_tidy() %>%
+    setNames(., make.names(names(.)))
 
   res <- stack_predict(object$equations[[type]], member_preds)