Change to quantile argument to quantile levels (#1208)

* quantile -> quantile_levels for #1203 * defer test until censored updates in new PR * update docs for quantile_levels * update test * disable quantile predictions for surv_reg --------- Co-authored-by: ‘topepo’ <‘[email protected]’>
tidymodels · Sep 26, 2024 · 861a64a · 861a64a
1 parent bef131b
commit 861a64a
Show file tree

Hide file tree

Showing 9 changed files with 43 additions and 61 deletions.
diff --git a/NEWS.md b/NEWS.md
@@ -11,6 +11,10 @@
 
 * New `extract_fit_time()` method has been added that returns the time it took to train the model (#853).
 
+## Breaking Change
+
+* For quantile prediction, the `predict()` argument has been changed from `quantile` to `quantile_levels` for consistency. This does not affect models with mode `"quantile regression"`. 
+* The quantile regression prediction type was disabled for the deprecated `surv_reg()` model. 
 
 # parsnip 1.2.1
 

diff --git a/R/predict.R b/R/predict.R
@@ -344,7 +344,7 @@ check_pred_type_dots <- function(object, type, ..., call = rlang::caller_env())
 
   # ----------------------------------------------------------------------------
 
-  other_args <- c("interval", "level", "std_error", "quantile",
+  other_args <- c("interval", "level", "std_error", "quantile_levels",
                   "time", "eval_time", "increasing")
   is_pred_arg <- names(the_dots) %in% other_args
   if (any(!is_pred_arg)) {

diff --git a/R/predict_quantile.R b/R/predict_quantile.R
@@ -1,14 +1,15 @@
 #' @keywords internal
 #' @rdname other_predict
-#' @param quantile A vector of numbers between 0 and 1 for the quantile being
-#'  predicted.
+#' @param quantile_levels  A vector of values between zero and one for the
+#' quantile to be predicted. If the model has a `"censored regression"` mode,
+#' this value should be `NULL`. For other modes, the default is `(1:9)/10`.
 #' @inheritParams predict.model_fit
 #' @method predict_quantile model_fit
 #' @export predict_quantile.model_fit
 #' @export
 predict_quantile.model_fit <- function(object,
                                        new_data,
-                                       quantile = (1:9)/10,
+                                       quantile_levels = NULL,
                                        interval = "none",
                                        level = 0.95,
                                        ...) {
@@ -20,15 +21,27 @@ predict_quantile.model_fit <- function(object,
     return(NULL)
   }
 
+  if (object$spec$mode == "quantile regression") {
+    if (!is.null(quantile_levels)) {
+      cli::cli_abort("When the mode is {.val quantile regression},
+                     {.arg quantile_levels} are specified by {.fn set_mode}.")
+    }
+  } else {
+    if (is.null(quantile_levels)) {
+      quantile_levels <- (1:9)/10
+    }
+    hardhat::check_quantile_levels(quantile_levels)
+    # Pass some extra arguments to be used in post-processor
+    object$quantile_levels <- quantile_levels
+  }
+
   new_data <- prepare_data(object, new_data)
 
   # preprocess data
   if (!is.null(object$spec$method$pred$quantile$pre)) {
     new_data <- object$spec$method$pred$quantile$pre(new_data, object)
   }
 
-  # Pass some extra arguments to be used in post-processor
-  object$spec$method$pred$quantile$args$p <- quantile
   pred_call <- make_pred_call(object$spec$method$pred$quantile)
 
   res <- eval_tidy(pred_call)

diff --git a/R/surv_reg_data.R b/R/surv_reg_data.R
@@ -59,25 +59,6 @@ set_pred(
   )
 )
 
-set_pred(
-  model = "surv_reg",
-  eng = "flexsurv",
-  mode = "regression",
-  type = "quantile",
-  value = list(
-    pre = NULL,
-    post = flexsurv_quant,
-    func = c(fun = "summary"),
-    args =
-      list(
-        object = expr(object$fit),
-        newdata = expr(new_data),
-        type = "quantile",
-        quantiles = expr(quantile)
-      )
-  )
-)
-
 # ------------------------------------------------------------------------------
 
 set_model_engine("surv_reg", mode = "regression", eng = "survival")
@@ -133,22 +114,3 @@ set_pred(
       )
   )
 )
-
-set_pred(
-  model = "surv_reg",
-  eng = "survival",
-  mode = "regression",
-  type = "quantile",
-  value = list(
-    pre = NULL,
-    post = survreg_quant,
-    func = c(fun = "predict"),
-    args =
-      list(
-        object = expr(object$fit),
-        newdata = expr(new_data),
-        type = "quantile",
-        p = expr(quantile)
-      )
-  )
-)
diff --git a/man/other_predict.Rd b/man/other_predict.Rd
diff --git a/man/set_args.Rd b/man/set_args.Rd
diff --git a/tests/testthat/_snaps/linear_reg_quantreg.md b/tests/testthat/_snaps/linear_reg_quantreg.md
@@ -0,0 +1,9 @@
+# linear quantile regression via quantreg - multiple quantiles
+
+    Code
+      ten_quant_pred <- predict(ten_quant, new_data = sac_test, quantile_levels = (0:
+      9) / 9)
+    Condition
+      Error in `predict_quantile()`:
+      ! When the mode is "quantile regression", `quantile_levels` are specified by `set_mode()`.
+
diff --git a/tests/testthat/test-linear_reg_quantreg.R b/tests/testthat/test-linear_reg_quantreg.R
@@ -83,6 +83,11 @@ test_that('linear quantile regression via quantreg - multiple quantiles', {
   expect_named(ten_quant_df, c(".pred_quantile", ".quantile_levels", ".row"))
   expect_true(nrow(ten_quant_df) == nrow(sac_test) * 10)
 
+  expect_snapshot(
+    ten_quant_pred <- predict(ten_quant, new_data = sac_test, quantile_levels = (0:9)/9),
+    error = TRUE
+  )
+
   ###
 
   ten_quant_one_row <- predict(ten_quant, new_data = sac_test[1,])

diff --git a/tests/testthat/test-surv_reg_survreg.R b/tests/testthat/test-surv_reg_survreg.R
@@ -10,8 +10,6 @@ complete_form <- survival::Surv(time) ~ group
 # ------------------------------------------------------------------------------
 
 test_that('survival execution', {
-  skip_on_travis()
-
   rlang::local_options(lifecycle_verbosity = "quiet")
   surv_basic <- surv_reg() %>% set_engine("survival")
   surv_lnorm <- surv_reg(dist = "lognormal") %>% set_engine("survival")
@@ -46,7 +44,7 @@ test_that('survival execution', {
 })
 
 test_that('survival prediction', {
-  skip_on_travis()
+  skip_if_not_installed("survival")
 
   rlang::local_options(lifecycle_verbosity = "quiet")
   surv_basic <- surv_reg() %>% set_engine("survival")
@@ -61,16 +59,6 @@ test_that('survival prediction', {
   exp_pred <- predict(extract_fit_engine(res), head(lung))
   exp_pred <- tibble(.pred = unname(exp_pred))
   expect_equal(exp_pred, predict(res, head(lung)))
-
-  exp_quant <- predict(extract_fit_engine(res), head(lung), p = (2:4)/5, type = "quantile")
-  exp_quant <-
-    apply(exp_quant, 1, function(x)
-      tibble(.pred = x, .quantile = (2:4) / 5))
-  exp_quant <- tibble(.pred = exp_quant)
-  obs_quant <- predict(res, head(lung), type = "quantile", quantile = (2:4)/5)
-
-  expect_equal(as.data.frame(exp_quant), as.data.frame(obs_quant))
-
 })