Error in dplyr's UseMethod("filter") : no applicable method for 'filter' applied to an object of class "NULL"

[yusufbtanriverdi]

The problem

I'm having trouble with predicting with a custom model created here tidymodels/tune#203, spesifically. I would like to use custom optimization techniques therefore I needed a custom tidymodel GPR. However, it is seen that the predict function cannot be achieved due to dplyr's filter. I am new to R but I searched on the internet for similar results and tried to comprehend where things could go wrong. I did some combinations in setting pred function with kernlab gpr to see if this is a mathematical issue rather, but the result did not change if not there was a incompatibility error.

I also proceeded with the same data in tidymodels/tune#203 but no luck.

Please notice I had to comment model translate code block and discard workflow as there were issues described.

Please tolerate any misformat or fault caused by me as I am only a newcomer to the R community.

Best,

Reproducible example

library(kernlab)
library(tidymodels)
library(conflicted)
library(dplyr)
library(tidymodels)
library(pso)
library(GA)
library(scales)

conflicts_prefer(kernlab::alpha)

tidymodels_prefer()
conflict_prefer("filter", "dplyr")

set_new_model("gauss_rbf")
set_model_mode(model = "gauss_rbf", mode = "regression")
set_model_engine("gauss_rbf", 
                 mode = "regression", 
                 eng = "kernlab")
set_dependency("gauss_rbf", eng = "kernlab", pkg = "kernlab")


set_model_arg(model = "gauss_rbf",
              eng = "kernlab",
              parsnip = "rbf_sigma", ## rbf_sigma exists for svm in dials
              original = "rbf_sigma",
              func = list(pkg = "dials", fun = "rbf_sigma"),
              has_submodel = FALSE)

gauss_rbf <- function(mode = "regression", rbf_sigma = NULL) {
  # Check for correct mode
  if (mode != "regression") {
    stop("`mode` should be 'regression'", call. = FALSE)
  }
  # Capture the arguments in quosures
  args <- list(rbf_sigma = rlang::enquo(rbf_sigma))
  # Save some empty slots for future parts of the specification
  out <- list(args = args, 
              eng_args = NULL,
              mode = mode, 
              method = NULL, 
              engine = NULL)
  # set classes in the correct order
  class(out) <- make_classes("gauss_rbf")
  out
}

gauss_rbf_wrap <- function(formula, data, rbf_sigma) {
  model <- kernlab::gausspr(x = formula, 
                            data = data, 
                            kpar = list(sigma = rbf_sigma),
                            type = "regression",
                            kernel = "rbfdot")
  return(model)
}

set_fit(model = "gauss_rbf",
        eng = "kernlab",
        mode = "regression",
        value = list(
          interface = "formula",
          protect = c("formula", "data"),
          func = c(fun = "gauss_rbf_wrap"),
          defaults = list()))


set_pred(model = "gauss_rbf", 
         eng   = "kernlab", 
         mode  = "regression", 
         type  = "numeric", 
         value = list(
           pre = NULL,
           post = function(results, object) as.vector(results) , 
           func = c(pkg = "kernlab", fun = "predict"), # need to register
           # the package, otherwise predict fails
           args = list(
             object = expr(object$fit), 
             newdata = expr(new_data), 
             type = "response")))

show_model_info("gauss_rbf")

#> Information for `gauss_rbf`
# modes: unknown, regression 
# 
# engines: 
#   regression: kernlab
# 
# ¹The model can use case weights.
# 
# arguments: 
#   kernlab: 
#   rbf_sigma --> rbf_sigma
# 
# fit modules:
#   engine       mode
# kernlab regression
# 
# prediction modules:
#   mode  engine methods
# regression kernlab numeric

gauss_model <- gauss_rbf(mode = "regression", rbf_sigma = 0.1) %>% 
  set_engine("kernlab") 

# ERROR : 
# Model fit template:
# Error in `call2()`:
#   ! `ns` must be a string
# Run `rlang::last_error()` to see where the error occurred.
# 
# gauss_model %>%
#   translate()


data("attrition")

attrition <- attrition %>%
  mutate(JobInvolvement = as.factor(JobInvolvement),
         Education = as.factor(Education),
         EnvironmentSatisfaction = as.factor(EnvironmentSatisfaction),
         PerformanceRating = as.factor(PerformanceRating),
         RelationshipSatisfaction = as.factor(RelationshipSatisfaction),
         WorkLifeBalance = as.factor(WorkLifeBalance),
         Attrition = factor(Attrition, levels = c("Yes", "No")))

set.seed(123)
intrain <- initial_split(attrition, prop = 0.8, strata = "Attrition")


rec <- recipe(Attrition ~ ., data = training(intrain)) %>% 
  themis::step_upsample(Attrition) %>% 
  step_scale(all_numeric()) %>% 
  step_nzv(all_numeric()) %>% 
  step_pca(all_numeric(), threshold = 0.9) %>% 
  step_rm(OverTime) %>% 
  prep()

training <- juice(rec)
testing <- bake(rec, testing(intrain))


training <- training %>%
  mutate(         Attrition = sapply(factor(Attrition, levels = c("Yes", "No")), unclass))

testing <- testing %>%
  mutate(         Attrition = sapply(factor(Attrition, levels = c("Yes", "No")), unclass))

gauss_fit <- gauss_model %>%
  fit(formula = Attrition ~., 
      data = training)

# gauss_fit
# 
# parsnip model object
# 
# Gaussian Processes object of class "gausspr" 
# Problem type: regression 
# 
# Gaussian Radial Basis kernel function. 
# Hyperparameter : sigma =  0.1 
# 
# Number of training instances learned : 1972 
# Train error : 0.151315008 

#ERROR: 
# Error in if (nrow(out) != 1L) { : argument is of length zero
# workflow() %>%
#   add_model(gauss_model) %>%
#   add_formula(Attrition ~ .) %>%
#   fit(data = training)


fit(gauss_model, Attrition ~ ., data = training)
# 
# parsnip model object
# 
# Gaussian Processes object of class "gausspr" 
# Problem type: regression 
# 
# Gaussian Radial Basis kernel function. 
# Hyperparameter : sigma =  0.1 
# 
# Number of training instances learned : 1972 
# Train error : 0.151315008 

predict(gauss_fit, 
        testing)

The output is:

Error in UseMethod("filter") :
no applicable method for 'filter' applied to an object of class "NULL"

The traceback:

> traceback()
9: dplyr::filter(., mode == object$spec$mode, engine == object$spec$engine)
8: dplyr::pull(., remove_intercept)
7: get_encoding(class(object$spec)[1]) %>% dplyr::filter(mode == 
       object$spec$mode, engine == object$spec$engine) %>% dplyr::pull(remove_intercept)
6: prepare_data(object, new_data)
5: predict_numeric.model_fit(object = object, new_data = new_data, 
       ...)
4: predict_numeric(object = object, new_data = new_data, ...)
3: predict.model_fit(gauss_fit, testing)
2: predict(gauss_fit, testing)
1: predict(gauss_fit, testing)

^{Created on 2023-03-08 with reprex v2.0.2}

Session info

sessioninfo::session_info()
#> ─ Session info ───────────────────────────────────────────────────────────────
#>  setting  value
#>  version  R version 4.2.2 (2022-10-31 ucrt)
#>  os       Windows 10 x64 (build 19044)
#>  system   x86_64, mingw32
#>  ui       RTerm
#>  language (EN)
#>  collate  English_United States.utf8
#>  ctype    English_United States.utf8
#>  tz       Europe/Paris
#>  date     2023-03-08
#>  pandoc   2.19.2 @ C:/Program Files/RStudio/resources/app/bin/quarto/bin/tools/ (via rmarkdown)
#> 
#> ─ Packages ───────────────────────────────────────────────────────────────────
#>  package     * version date (UTC) lib source
#>  cli           3.6.0   2023-01-09 [1] CRAN (R 4.2.2)
#>  clipr         0.8.0   2022-02-22 [1] CRAN (R 4.2.2)
#>  digest        0.6.31  2022-12-11 [1] CRAN (R 4.2.2)
#>  evaluate      0.20    2023-01-17 [1] CRAN (R 4.2.2)
#>  fastmap       1.1.1   2023-02-24 [1] CRAN (R 4.2.2)
#>  fs            1.6.1   2023-02-06 [1] CRAN (R 4.2.2)
#>  glue          1.6.2   2022-02-24 [1] CRAN (R 4.2.2)
#>  htmltools     0.5.4   2022-12-07 [1] CRAN (R 4.2.2)
#>  knitr         1.42    2023-01-25 [1] CRAN (R 4.2.2)
#>  lifecycle     1.0.3   2022-10-07 [1] CRAN (R 4.2.2)
#>  reprex        2.0.2   2022-08-17 [1] CRAN (R 4.2.2)
#>  rlang         1.0.6   2022-09-24 [1] CRAN (R 4.2.2)
#>  rmarkdown     2.20    2023-01-19 [1] CRAN (R 4.2.2)
#>  rstudioapi    0.14    2022-08-22 [1] CRAN (R 4.2.2)
#>  sessioninfo   1.2.2   2021-12-06 [1] CRAN (R 4.2.2)
#>  withr         2.5.0   2022-03-03 [1] CRAN (R 4.2.2)
#>  xfun          0.37    2023-01-31 [1] CRAN (R 4.2.2)
#>  yaml          2.3.7   2023-01-23 [1] CRAN (R 4.2.2)
#> 
#>  [1] C:/Users/yusuf/AppData/Local/R/win-library/4.2
#>  [2] C:/Program Files/R/R-4.2.2/library
#> 
#> ──────────────────────────────────────────────────────────────────────────────

[EmilHvitfeldt]

Hello @yusuftengriverdi 👋

You are missing a set_encoding() call, see https://www.tidymodels.org/learn/develop/models/ for more detail, adding the following will remove the error:

set_encoding(
  model = "gauss_rbf", 
  eng   = "kernlab", 
  mode  = "regression",
  options = list(
    predictor_indicators = "traditional",
    compute_intercept = TRUE,
    remove_intercept = TRUE,
    allow_sparse_x = FALSE
  )
)

[EmilHvitfeldt]

I think this isn't the right repo. But it might be nice to have better error message on this problem

[yusufbtanriverdi]

set_encoding(
  model = "gauss_rbf", 
  eng   = "kernlab", 
  mode  = "regression",
  options = list(
    predictor_indicators = "traditional",
    compute_intercept = TRUE,
    remove_intercept = TRUE,
    allow_sparse_x = FALSE
  )
)

Perfectly worked now! Appreciate the help. The error was indeed very distracting.

[EmilHvitfeldt]

Perfect! @yusuftengriverdi, i reopened the issue to remind us to create a better error message

[simonpcouch]

I agree with Emil—will go ahead and move this to parsnip, where it more likely ought to live.