Introduction

The {tidyAML} package is an R package that provides a set of tools for building regression/classification models on the fly with minimal input required. In this post we will discuss the create_workflow_set() function.

The create_workflow_set function is a function in the tidyAML package that is used to create a workflowset object from the workflowsets package. A workflow is a sequence of tasks that can be executed in a specific order, and is often used in data analysis and machine learning to automate data processing and model fitting. The create_workflow_set function takes as input a YAML specification of a set of workflows, and returns a list of workflow objects that can be executed using the tidymodels package and its associated packages.

The create_workflow_set function is particularly useful when working with the tidymodels package and the parsnip framework. The tidymodels package is a collection of packages for modeling and machine learning in R that provides a consistent interface for building, tuning, and evaluating machine learning models. The parsnip package is part of the tidymodels ecosystem and provides a way to specify a wide range of models in a consistent manner.

Function

Let’s take a look at the function and it’s arguments.

create_workflow_set(
  .model_tbl = NULL, 
  .recipe_list = list(), 
  .cross = TRUE
)

• .model_tbl – The model table that is generated from a function like fast_regression_parsnip_spec_tbl(). The model spec column will be grabbed automatically as the class of the object must be tidyaml_base_tbl
• .recipe_list – Provide a list of recipes here that will get added to the workflow set object.
• .cross – The default is TRUE, can be set to FALSE. This is passed to the cross parameter as an argument to the workflow_set() function.

Example

Here is a simple example. Remember you really only want to use this if you have a model_spec tibble not a tibble with workflows that have already been fit.

library(tidyAML)
library(recipes)

rec_obj <- recipe(mpg ~ ., data = mtcars)
spec_tbl <- fast_regression_parsnip_spec_tbl(
  .parsnip_fns = "linear_reg",
  .parsnip_eng = c("lm","glm")
)

wfs_tbl <- create_workflow_set(
  spec_tbl,
  list(rec_obj)
)

Now let’s inspect.

library(dplyr)

wfs_tbl |>
  slice(1)

# A workflow set/tibble: 1 × 4
  wflow_id            info             option    result    
  <chr>               <list>           <list>    <list>    
1 recipe_linear_reg_1 <tibble [1 × 4]> <opts[0]> <list [0]>

class(wfs_tbl)

[1] "workflow_set" "tbl_df"       "tbl"          "data.frame"  

wfs_tbl$info

[[1]]
# A tibble: 1 × 4
  workflow   preproc model      comment
  <list>     <chr>   <chr>      <chr>  
1 <workflow> recipe  linear_reg ""     

[[2]]
# A tibble: 1 × 4
  workflow   preproc model      comment
  <list>     <chr>   <chr>      <chr>  
1 <workflow> recipe  linear_reg ""     

wfs_tbl$info[[1]]$workflow[[1]]

══ Workflow ════════════════════════════════════════════════════════════════════
Preprocessor: Recipe
Model: linear_reg()

── Preprocessor ────────────────────────────────────────────────────────────────
0 Recipe Steps

── Model ───────────────────────────────────────────────────────────────────────
Linear Regression Model Specification (regression)

Computational engine: lm 

Voila!