Parralel processing fails with function \"all_nominal\" (related to to issue #159)

[kelseygonzalez]

Issued discovered during the Applied machine learning workshop. Related to issue #159 .
suggested to submit issue from @jyuu .

library(tidymodels)
#> Registered S3 method overwritten by 'xts':
#>   method     from
#>   as.zoo.xts zoo
#> -- Attaching packages ----------------------------------------------------------- tidymodels 0.0.3 --
#> v broom     0.5.3     v purrr     0.3.3
#> v dials     0.0.4     v recipes   0.1.9
#> v dplyr     0.8.3     v rsample   0.0.5
#> v ggplot2   3.2.1     v tibble    2.1.3
#> v infer     0.5.1     v yardstick 0.0.4
#> v parsnip   0.0.5
#> -- Conflicts -------------------------------------------------------------- tidymodels_conflicts() --
#> x purrr::discard()    masks scales::discard()
#> x dplyr::filter()     masks stats::filter()
#> x dplyr::lag()        masks stats::lag()
#> x ggplot2::margin()   masks dials::margin()
#> x recipes::step()     masks stats::step()
#> x recipes::yj_trans() masks scales::yj_trans()
library(tune)
library(doParallel)
#> Loading required package: foreach
#> 
#> Attaching package: 'foreach'
#> The following objects are masked from 'package:purrr':
#> 
#>     accumulate, when
#> Loading required package: iterators
#> Loading required package: parallel

data(Chicago)

us_hol <- 
  timeDate::listHolidays() %>% 
  stringr::str_subset("(^US)|(Easter)")

chi_rec <-
  recipe(ridership ~ ., data = Chicago) %>%
  step_holiday(date, holidays = us_hol) %>%
  step_date(date) %>%
  step_rm(date) %>%
  step_dummy(all_nominal()) %>%
  step_zv(all_predictors())

chi_folds <- rolling_origin(Chicago, initial = 364 * 15, assess = 7 * 4, skip = 7 * 4, cumulative = FALSE)

glmn_grid <- expand.grid(penalty = 10^seq(-3, -1, 
                                          length.out = 20), 
                         mixture = (0:5)/5)

glmn_rec <- chi_rec %>%  step_normalize(all_predictors()) 
glmn_mod <-
  linear_reg(penalty = tune(), mixture = tune()) %>% set_engine("glmnet")

ctrl <- control_grid(save_pred = TRUE,
                           verbose = TRUE)

glmn_tune <-
  tune_grid(
    glmn_rec,
    model = glmn_mod,
    resamples = chi_folds,
    grid = glmn_grid,
    control = ctrl
  )
#> i Slice1: recipe
#> v Slice1: recipe
#> i Slice1: model 1/6
#> v Slice1: model 1/6
#> i Slice1: model 1/6 (predictions)
#> i Slice1: model 2/6
#> v Slice1: model 2/6
#> i Slice1: model 2/6 (predictions)
#> i Slice1: model 3/6
#> v Slice1: model 3/6
#> i Slice1: model 3/6 (predictions)
#> i Slice1: model 4/6
#> v Slice1: model 4/6
#> i Slice1: model 4/6 (predictions)
#> i Slice1: model 5/6
#> v Slice1: model 5/6
#> i Slice1: model 5/6 (predictions)
#> i Slice1: model 6/6
#> v Slice1: model 6/6
#> i Slice1: model 6/6 (predictions)
#> i Slice2: recipe
#> v Slice2: recipe
#> i Slice2: model 1/6
#> v Slice2: model 1/6
#> i Slice2: model 1/6 (predictions)
#> i Slice2: model 2/6
#> v Slice2: model 2/6
#> i Slice2: model 2/6 (predictions)
#> i Slice2: model 3/6
#> v Slice2: model 3/6
#> i Slice2: model 3/6 (predictions)
#> i Slice2: model 4/6
#> v Slice2: model 4/6
#> i Slice2: model 4/6 (predictions)
#> i Slice2: model 5/6
#> v Slice2: model 5/6
#> i Slice2: model 5/6 (predictions)
#> i Slice2: model 6/6
#> v Slice2: model 6/6
#> i Slice2: model 6/6 (predictions)
#> i Slice3: recipe
#> v Slice3: recipe
#> i Slice3: model 1/6
#> v Slice3: model 1/6
#> i Slice3: model 1/6 (predictions)
#> i Slice3: model 2/6
#> v Slice3: model 2/6
#> i Slice3: model 2/6 (predictions)
#> i Slice3: model 3/6
#> v Slice3: model 3/6
#> i Slice3: model 3/6 (predictions)
#> i Slice3: model 4/6
#> v Slice3: model 4/6
#> i Slice3: model 4/6 (predictions)
#> i Slice3: model 5/6
#> v Slice3: model 5/6
#> i Slice3: model 5/6 (predictions)
#> i Slice3: model 6/6
#> v Slice3: model 6/6
#> i Slice3: model 6/6 (predictions)
#> i Slice4: recipe
#> v Slice4: recipe
#> i Slice4: model 1/6
#> v Slice4: model 1/6
#> i Slice4: model 1/6 (predictions)
#> i Slice4: model 2/6
#> v Slice4: model 2/6
#> i Slice4: model 2/6 (predictions)
#> i Slice4: model 3/6
#> v Slice4: model 3/6
#> i Slice4: model 3/6 (predictions)
#> i Slice4: model 4/6
#> v Slice4: model 4/6
#> i Slice4: model 4/6 (predictions)
#> i Slice4: model 5/6
#> v Slice4: model 5/6
#> i Slice4: model 5/6 (predictions)
#> i Slice4: model 6/6
#> v Slice4: model 6/6
#> i Slice4: model 6/6 (predictions)
#> i Slice5: recipe
#> v Slice5: recipe
#> i Slice5: model 1/6
#> v Slice5: model 1/6
#> i Slice5: model 1/6 (predictions)
#> i Slice5: model 2/6
#> v Slice5: model 2/6
#> i Slice5: model 2/6 (predictions)
#> i Slice5: model 3/6
#> v Slice5: model 3/6
#> i Slice5: model 3/6 (predictions)
#> i Slice5: model 4/6
#> v Slice5: model 4/6
#> i Slice5: model 4/6 (predictions)
#> i Slice5: model 5/6
#> v Slice5: model 5/6
#> i Slice5: model 5/6 (predictions)
#> i Slice5: model 6/6
#> v Slice5: model 6/6
#> i Slice5: model 6/6 (predictions)
#> i Slice6: recipe
#> v Slice6: recipe
#> i Slice6: model 1/6
#> v Slice6: model 1/6
#> i Slice6: model 1/6 (predictions)
#> i Slice6: model 2/6
#> v Slice6: model 2/6
#> i Slice6: model 2/6 (predictions)
#> i Slice6: model 3/6
#> v Slice6: model 3/6
#> i Slice6: model 3/6 (predictions)
#> i Slice6: model 4/6
#> v Slice6: model 4/6
#> i Slice6: model 4/6 (predictions)
#> i Slice6: model 5/6
#> v Slice6: model 5/6
#> i Slice6: model 5/6 (predictions)
#> i Slice6: model 6/6
#> v Slice6: model 6/6
#> i Slice6: model 6/6 (predictions)
#> i Slice7: recipe
#> v Slice7: recipe
#> i Slice7: model 1/6
#> v Slice7: model 1/6
#> i Slice7: model 1/6 (predictions)
#> i Slice7: model 2/6
#> v Slice7: model 2/6
#> i Slice7: model 2/6 (predictions)
#> i Slice7: model 3/6
#> v Slice7: model 3/6
#> i Slice7: model 3/6 (predictions)
#> i Slice7: model 4/6
#> v Slice7: model 4/6
#> i Slice7: model 4/6 (predictions)
#> i Slice7: model 5/6
#> v Slice7: model 5/6
#> i Slice7: model 5/6 (predictions)
#> i Slice7: model 6/6
#> v Slice7: model 6/6
#> i Slice7: model 6/6 (predictions)
#> i Slice8: recipe
#> v Slice8: recipe
#> i Slice8: model 1/6
#> v Slice8: model 1/6
#> i Slice8: model 1/6 (predictions)
#> i Slice8: model 2/6
#> v Slice8: model 2/6
#> i Slice8: model 2/6 (predictions)
#> i Slice8: model 3/6
#> v Slice8: model 3/6
#> i Slice8: model 3/6 (predictions)
#> i Slice8: model 4/6
#> v Slice8: model 4/6
#> i Slice8: model 4/6 (predictions)
#> i Slice8: model 5/6
#> v Slice8: model 5/6
#> i Slice8: model 5/6 (predictions)
#> i Slice8: model 6/6
#> v Slice8: model 6/6
#> i Slice8: model 6/6 (predictions)
# this works fine, but in parralel processing....


# parallel processing... 

parallel::detectCores(logical = FALSE)
#> [1] 4
# = 4 cores on my computer

registerDoParallel(makeCluster(4))

# run `tune_grid()`...
glmn_tune <-
  tune::tune_grid(
    glmn_rec,
    model = glmn_mod,
    resamples = chi_folds,
    grid = glmn_grid,
    control = ctrl
  )
#> Warning: All models failed in tune_grid(). See the `.notes` column.

glmn_tune$.notes[[1]]
#> # A tibble: 1 x 1
#>   .notes                                                                   
#>   <chr>                                                                    
#> 1 "recipe: Error in all_nominal(): could not find function \"all_nominal\""

^{Created on 2020-01-28 by the reprex package (v0.3.0)}

[maxheld83]

I think this too, might be a duplicate of #157, or have the same root cause.
See comment in #157.

[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.