Soft deprecate liquidSVM engine

DESCRIPTION

@@ -24,6 +24,7 @@ Imports:
     tibble (>= 2.1.1),
     generics (>= 0.1.0),
     glue,
+    lifecycle,
     magrittr,
     stats,
     tidyr (>= 1.0.0),
@@ -51,5 +52,4 @@ Suggests:
     MASS,
     nlme,
     modeldata,
-    liquidSVM,
     Matrix

NEWS.md

@@ -1,6 +1,10 @@
+# parsnip (development version)
+
+* The `liquidSVM` engine for `svm_rbf()` was deprecated due to that package's removal from CRAN.
+
 # parsnip 0.1.5
 
-* An RStudio add-in is availble that makes writing multiple `parsnip` model specifications to the source window. It can be accessed via the IDE addin menus or by calling `parsnip_addin()`.
+* An RStudio add-in is available that makes writing multiple `parsnip` model specifications to the source window. It can be accessed via the IDE addin menus or by calling `parsnip_addin()`.
 
 * For `xgboost` models, users can now pass `objective` to `set_engine("xgboost")`. (#403)
 

R/engines.R

@@ -93,6 +93,12 @@ set_engine <- function(object, engine, ...) {
   }
   if (!is.character(engine) | length(engine) != 1)
     rlang::abort("`engine` should be a single character value.")
+  if (engine == "liquidSVM") {
+    lifecycle::deprecate_soft(
+      "0.1.6",
+      "set_engine(engine = 'cannot be liquidSVM')",
+      details = "The liquidSVM package is no longer available on CRAN.")
+  }
 
   object$engine <- engine
   object <- check_engine(object)

R/svm_rbf.R

@@ -33,7 +33,6 @@
 #'  following _engines_:
 #' \itemize{
 #' \item \pkg{R}:  `"kernlab"` (the default)
-#' \item \pkg{R}:  `"liquidSVM"`
 #' }
 #'
 #'

man/rmd/svm-rbf.Rmd

@@ -24,34 +24,6 @@ svm_rbf() %>%
 
 `fit()` passes the data directly to `kernlab::ksvm()` so that its formula method can create dummy variables as-needed. 
 
-## liquidSVM
-
-```{r liquidSVM-reg}
-svm_rbf() %>% 
-  set_engine("liquidSVM") %>% 
-  set_mode("regression") %>% 
-  translate()
-```
-
-```{r liquidSVM-cls}
-svm_rbf() %>% 
-  set_engine("liquidSVM") %>% 
-  set_mode("classification") %>% 
-  translate()
-```
-
-Note that models created using the `liquidSVM` engine cannot be saved like
-conventional R objects. The `fit` slot of the `model_fit` object has to be saved
-separately using the `liquidSVM::write.liquidSVM()` function. Likewise to restore
-a model, the `fit` slot has to be replaced with the model that is read using the
-`liquidSVM::read.liquidSVM()` function.
-
-`liquidSVM` parameterizes the kernel parameter differently than `kernlab`. To 
-translate between engines, `sigma = 1/gammas^2`. Users will be specifying 
-`sigma` and the function translates the value to `gamma`. 
-
-`fit()` passes the data directly to `liquidSVM::svm()` so that its formula method can create dummy variables as-needed. 
-
 ## Parameter translations
 
 The standardized parameter names in parsnip can be mapped to their original 
@@ -66,8 +38,6 @@ get_defaults_svm_rbf <- function() {
     "svm_rbf", "kernlab",    "cost",          "C",        "1",
     "svm_rbf", "kernlab",    "rbf_sigma",     "sigma",    "varies",
     "svm_rbf", "kernlab",    "margin",        "epsilon",  "0.1",
-    "svm_rbf", "liquidSVM",  "cost",          "lambdas",  "varies",
-    "svm_rbf", "liquidSVM",  "rbf_sigma",     "gammas",   "varies",
   )
 }
 convert_args("svm_rbf")

tests/testthat/test_svm_liquidsvm.R

@@ -79,132 +79,3 @@ test_that('bad input', {
   expect_error(svm_rbf(mode = "reallyunknown"))
   expect_error(translate(svm_rbf() %>% set_engine( NULL)))
 })
-
-# ------------------------------------------------------------------------------
-# define model specification for classification and regression
-
-reg_mod <-
-  svm_rbf(rbf_sigma = .1, cost = 0.25) %>%
-  set_engine("liquidSVM", random_seed = 1234, folds = 1, max_gamma=500) %>%
-  set_mode("regression")
-
-cls_mod <-
-  svm_rbf(rbf_sigma = .1, cost = 0.125) %>%
-  set_engine("liquidSVM", random_seed = 1234, folds = 1) %>%
-  set_mode("classification")
-
-ctrl <- fit_control(verbosity = 0, catch = FALSE)
-
-# ------------------------------------------------------------------------------
-
-test_that('svm rbf regression', {
-
-  skip_if_not_installed("liquidSVM")
-
-  expect_warning(
-    expect_error(
-      fit_xy(
-        reg_mod,
-        control = ctrl,
-        x = hpc[, 2:4],
-        y = hpc$compounds
-      ),
-      regexp = NA
-    )
-  )
-
-  expect_warning(
-    expect_error(
-      fit(
-        reg_mod,
-        compounds ~ .,
-        data = hpc[, -5],
-        control = ctrl
-      ),
-      regexp = NA
-    )
-  )
-
-})
-
-
-test_that('svm rbf regression prediction', {
-
-  skip_if_not_installed("liquidSVM")
-
-  expect_warning(
-    reg_form <-
-      fit(
-        object = reg_mod,
-        formula = compounds ~ .,
-        data = hpc[, -5],
-        control = ctrl
-      )
-  )
-
-  expect_warning(
-    reg_xy_form <-
-      fit_xy(
-        object = reg_mod,
-        x = hpc[, 2:4],
-        y = hpc$compounds,
-        control = ctrl
-      )
-  )
-  expect_equal(reg_form$spec, reg_xy_form$spec)
-
-  expect_warning(
-    liquidSVM_form <-
-      liquidSVM::svm(
-        x = compounds ~ .,
-        y = hpc[, -5],
-        gammas = .1,
-        lambdas = 0.25,
-        folds = 1,
-        random_seed = 1234
-      )
-  )
-
-  expect_warning(
-    liquidSVM_xy_form <-
-      liquidSVM::svm(
-        x = hpc[, 2:4],
-        y = hpc$compounds,
-        gammas = .1,
-        lambdas = 0.25,
-        folds = 1,
-        random_seed = 1234
-      )
-  )
-
-  # check coeffs for liquidSVM formula and liquidSVM xy fit interfaces
-  expect_equal(liquidSVM::getSolution(liquidSVM_form)[c("coeff", "sv")],
-               liquidSVM::getSolution(liquidSVM_xy_form)[c("coeff", "sv")])
-
-  # check predictions for liquidSVM formula and liquidSVM xy interfaces
-  liquidSVM_form_preds <- predict(liquidSVM_form, hpc[1:3, 2:4])
-  liquidSVM_form_xy_preds <- predict(liquidSVM_xy_form, hpc[1:3, 2:4])
-  expect_equal(liquidSVM_form_preds, liquidSVM_form_xy_preds)
-
-  # check predictions for parsnip formula and liquidSVM formula interfaces
-  liquidSVM_pred <-
-    structure(
-      list(.pred = liquidSVM_form_preds),
-      row.names = c(NA, -3L), class = c("tbl_df", "tbl", "data.frame"))
-
-  parsnip_pred <- predict(reg_form, hpc[1:3, 2:4])
-  expect_equal(as.data.frame(liquidSVM_pred), as.data.frame(parsnip_pred))
-
-  # check that coeffs are equal for formula methods called via parsnip and liquidSVM
-  expect_equal(liquidSVM::getSolution(reg_form$fit)[c("coeff", "sv")],
-               liquidSVM::getSolution(liquidSVM_form)[c("coeff", "sv")])
-
-  # check coeffs are equivalent for parsnip fit_xy and parsnip formula methods
-  expect_equal(liquidSVM::getSolution(reg_form$fit)[c("coeff", "sv")],
-               liquidSVM::getSolution(reg_xy_form$fit)[c("coeff", "sv")])
-
-  # check predictions are equal for parsnip xy and liquidSVM xy methods
-  parsnip_xy_pred <- predict(reg_xy_form, hpc[1:3, -c(1, 5)])
-  expect_equal(as.data.frame(liquidSVM_pred), as.data.frame(parsnip_xy_pred))
-})
-