Ranger not working with sparse representation

[tolliam]

Ranger models fit without any problems when using a sparse representation but produce an error about converting matrices to data frames when generating predictions. I guess this relates to when the predictions are being converted back from ranger's output? With issue 691 I think this means only glmnet is the only engine supported for sparse matrices.

library(tidyverse)
library(tidymodels)
#> Warning: package 'dials' was built under R version 4.1.2
#> Warning: package 'recipes' was built under R version 4.1.2
#> Warning: package 'workflowsets' was built under R version 4.1.2

data("small_fine_foods")
training_data
#> # A tibble: 4,000 × 3
#>    product    review                                                       score
#>    <chr>      <chr>                                                        <fct>
#>  1 B000J0LSBG "this stuff is  not stuffing  its  not good at all  save yo… other
#>  2 B000EYLDYE "I absolutely LOVE this dried fruit.  LOVE IT.  Whenever I … great
#>  3 B0026LIO9A "GREAT DEAL, CONVENIENT TOO.  Much cheaper than WalMart and… great
#>  4 B00473P8SK "Great flavor, we go through a ton of this sauce! I discove… great
#>  5 B001SAWTNM "This is excellent salsa/hot sauce, but you can get it for … great
#>  6 B000FAG90U "Again, this is the best dogfood out there.  One suggestion… great
#>  7 B006BXTCEK "The box I received was filled with teas, hot chocolates, a… other
#>  8 B002GWH5OY "This is delicious coffee which compares favorably with muc… great
#>  9 B003R0MFYY "Don't let these little tiny cans fool you.  They pack a lo… great
#> 10 B001EO5ZXI "One of the nicest, smoothest cup of chai I've made. Nice m… great
#> # … with 3,990 more rows

library(hardhat)
#> Warning: package 'hardhat' was built under R version 4.1.2
sparse_bp <- default_recipe_blueprint(composition = "dgCMatrix")

library(textrecipes)

text_rec <-
  recipe(score ~ review, data = training_data) %>%
  step_tokenize(review)  %>%
  step_tokenfilter(review, max_tokens = 1e3) %>%
  step_tfidf(review)

ranger_spec <-
  rand_forest(mode = "classification") %>%
  set_engine("ranger")

wf_sparse <- 
  workflow() %>%
  add_recipe(text_rec, blueprint = sparse_bp) %>%
  add_model(ranger_spec)

wf_default <- 
  workflow() %>%
  add_recipe(text_rec) %>%
  add_model(ranger_spec)

set.seed(123)

wf_default %>% 
  fit(training_data) %>% 
  predict(testing_data)
#> # A tibble: 1,000 × 1
#>    .pred_class
#>    <fct>      
#>  1 great      
#>  2 great      
#>  3 other      
#>  4 great      
#>  5 great      
#>  6 great      
#>  7 great      
#>  8 great      
#>  9 great      
#> 10 other      
#> # … with 990 more rows

wf_sparse %>% 
  fit(training_data) %>% 
  predict(testing_data)
#> Error in as.data.frame.default(new_data): cannot coerce class 'structure("dgCMatrix", package = "Matrix")' to a data.frame

^{Created on 2022-03-26 by the reprex package (v2.0.1)}

[juliasilge]

For clarity and anybody else coming by, we'd like to clarify that tidymodels fully supports using a sparse matrix representation with xgboost. However, xgboost itself behaves differently with sparse data compared to a dense representation of the same data.

Here is a smaller reprex for the predicting problem with ranger:

library(tidymodels)
data(two_class_dat)

data_train <- two_class_dat[-(1:10), ]
data_test  <- two_class_dat[  1:10 , ]

library(hardhat)
sparse_bp <- default_recipe_blueprint(composition = "dgCMatrix")

simple_rec <- recipe(Class ~ ., data = data_train)

rf_spec <-
  rand_forest(mode = "classification") %>%
  set_engine("ranger")

wf <- 
  workflow() %>%
  add_recipe(simple_rec, blueprint = sparse_bp) %>%
  add_model(rf_spec) %>%
  fit(data = data_train)

predict(wf, data_test)
#> Error in as.data.frame.default(new_data): cannot coerce class 'structure("dgCMatrix", package = "Matrix")' to a data.frame

^{Created on 2022-03-29 by the reprex package (v2.0.1)}

[hfrick]

closed by #950

[github-actions]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.