Why can't I fit my model with the tidymodels package

[zhangkai]

I recently saw a textbook about the tidymodels package. I want to use the colon dataset in the survival package for testing, but I don't know why I encountered a problem.

My code is as follows

library(tidymodels)
library(survival)

data(colon)
str(colon)

colon$sex <- ifelse(colon$sex==1,"male","female")
colon$obstruct <- ifelse(colon$obstruct ==1,"yes","no")
colon$perfor <- ifelse(colon$perfor ==1,"yes","no")
colon$adhere <- ifelse(colon$adhere ==1,"yes","no")
colon$status <- ifelse(colon$obstruct ==1,"death","alive")
colon$node4<- ifelse(colon$node4 ==1,"yes","no")


colon <- select(colon,id,age,rx,sex,age,obstruct,perfor,adhere,nodes,status)
colon <- na.omit(colon)

data_split<- initial_split(colon,
                           prop = 3/4,
                           strata = status)

train_data <- training(data_split)
test_data<- testing(data_split)

str(train_data)
train_rec <-
  recipe(status ~., data = train_data) %>%
  update_role(id, new_role = "ID")%>%
  step_zv(all_numeric(),-all_outcomes()) %>%
  step_normalize(all_numeric(),-all_outcomes())%>%
  step_novel(all_nominal(),-all_outcomes()) %>%
  step_dummy(all_nominal(),-all_outcomes())

 

summary(train_rec)

prepped_data <-
  train_rec %>% # use the recipe object
  prep() %>% # perform the recipe on training data
  juice() # extract only the preprocessed dataframe

glimpse(prepped_data)

set.seed(100)

cv_folds <-
  vfold_cv(train_data,
           v = 5,
           strata = status)


log_spec <- # your model specification
  logistic_reg() %>%  # model type
  set_engine(engine = "glm") %>%  # model engine
  set_mode("classification") # model mode

log_wflow <- # new workflow object
  workflow() %>% # use workflow function
  add_recipe(train_rec) %>%   # use the new recipe
  add_model(log_spec)   # add your model spec

log_res <-
  log_wflow  %>%
  fit_resamples(
    resamples = cv_folds,
    metrics = metric_set(
      precision, f_meas,
      accuracy, kap,
      roc_auc, sens, spec),
    control = control_resamples(
      save_pred = TRUE)
  )

log_res$.notes

Is there a few things I don't quite understand, or is it a bug in this package?

1. Why does my model not fit? I think it may be a problem with the recipe step.
2. I already have the ID variable defined, why does step_normalize also normalize it?
3. For the binary variable of gender, how does the sex_new that appears after dummy explain?

[Max]

There are two issues.

First, I think that you have a copy/paste issue. The status conversions should be

colon$status <- ifelse(colon$status ==1,"death","alive")

After that, I believe that we have a bug related to using the new role for id. If you take that out of the data and recipe, it works.

I filed a bug to figure this out.

[zhangkai]

Thanks a lot for your answer.
But I don't know how to solve the third problem

1. For the binary variable of gender, how does the sex_new that appears after dummy explain?
   My current solution is to put all_numeric_predictors() at the end, what is the purpose of generating this useless variable?
   
   

[Max]

You used the recipe step:

@zhangkai

step_novel(all_nominal(),-all_outcomes()) %>%

That adds a new factor level.

[zhangkai]

thanks a lot for your answer