Don't coerce sparse data to non-sparse during predict() (#950)

* pass sparse data through prepare_data() * add news * move news bullet to devel --------- Co-authored-by: Max Kuhn <[email protected]>
tidymodels · May 17, 2023 · 145bac2 · 145bac2
1 parent fe721f6
commit 145bac2
Show file tree

Hide file tree

Showing 2 changed files with 6 additions and 0 deletions.
diff --git a/NEWS.md b/NEWS.md
@@ -1,5 +1,7 @@
 # parsnip (development version)
 
+* Fixed bug where sparse data was being coerced to non-sparse format doing `predict()`. 
+
 * `augment()` now works for censored regression models. 
 
 # parsnip 1.1.0

diff --git a/R/predict.R b/R/predict.R
@@ -426,6 +426,10 @@ prepare_data <- function(object, new_data) {
     new_data <- new_data[, colnames(new_data) != "(Intercept)", drop = FALSE]
   }
 
+  if (allow_sparse(object) && inherits(new_data, "dgCMatrix")) {
+    return(new_data)
+  }
+
   fit_interface <- object$spec$method$fit$interface
   switch(
     fit_interface,