The building of {tidyAML}

[This article was first published on Steve's Data Tips and Tricks, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

Introduction

Yesterday I posted on An Update to {tidyAML} where I was discussing some of my thought process and how things could potentially work for the package.

Today I want to showcase how the function fast_regression_parsnip_spec_tbl() and it’s complimentary function fast_classification_parsnip_spec_tbl() actually work or maybe don’t work for that matter.

We are going to pick on fast_regression_parsnip_spec_tbl() in today’s post. The point of it is that it creates a tibble of parsnip regression model specifications. This will create a tibble of 46 different regression model specifications which can be filtered. The model specs are created first and then filtered out. This will only create models for regression problems. To find all of the supported models in this package you can visit the parsnip search page

Function

First let’s take a look at the function call itself.

fast_regression_parsnip_spec_tbl(
  .parsnip_fns = "all", 
  .parsnip_eng = "all"
  )

Now let’s take a look at the arguments:

  • .parsnip_fns – The default for this is set to all. This means that all of the parsnip linear regression functions will be used, for example linear_reg(), or cubist_rules. You can also choose to pass a c() vector like c("linear_reg","cubist_rules")
  • .parsnip_eng – The default for this is set to all. This means that all of the parsnip linear regression engines will be used, for example lm, or glm. You can also choose to pass a c() vector like c('lm', 'glm')

The workhorse to this function is the internal_make_spec_tbl() function. This is the one that will be the subject of the post. Let’s take a look at it’s inner workings, afterall this is open source.

internal_make_spec_tbl <- function(.data){

  # Checks ----
  df <- dplyr::as_tibble(.data)

  nms <- unique(names(df))

  if (!".parsnip_engine" %in% nms | !".parsnip_mode" %in% nms | !".parsnip_fns" %in% nms){
    rlang::abort(
      message = "The model tibble must come from the class/reg to parsnip function.",
      use_cli_format = TRUE
    )
  }

  # Make tibble ----
  mod_spec_tbl <- df %>%
    dplyr::mutate(
      model_spec = purrr::pmap(
        dplyr::cur_data(),
        ~ match.fun(..3)(mode = ..2, engine = ..1)
      )
    ) %>%
    # add .model_id column
    dplyr::mutate(.model_id = dplyr::row_number()) %>%
    dplyr::select(.model_id, dplyr::everything())

  # Return ----
  return(mod_spec_tbl)

}

Let’s examine this (and it is currently changing form in a github issue). Firstly, we are taking in a data.frame/tibble that has to have certain names in it (this is going to change and look for a class instead). Once this determination is TRUE we then proceed to the meat and potatoes of it. The internal mod_spec_tbl is made using mutate, pmap, cur_data and match.fun. What this does essentially is the following:

  1. mutate a column called model_spec
  2. Use the {purrr} function pmap which maps over several columns in parallel to create the model spec.
  3. Inside of the pmap we use cur_data() to get the current line where we match the function using match.fun (which takes a character string of the function, this means the library needs to be loaded) we supply the column it is in and then we supply the arguments we want.
  4. We give it a numeric model id
  5. We then ensure that the .model_id column is first.

Example

Let’s see it in action!

library(tidyAML) # Not yet available, you can install from GitHub though

fast_regression_parsnip_spec_tbl()
# A tibble: 46 × 5
   .model_id .parsnip_engine .parsnip_mode .parsnip_fns model_spec
       <int> <chr>           <chr>         <chr>        <list>    
 1         1 lm              regression    linear_reg   <spec[+]> 
 2         2 brulee          regression    linear_reg   <spec[+]> 
 3         3 gee             regression    linear_reg   <spec[+]> 
 4         4 glm             regression    linear_reg   <spec[+]> 
 5         5 glmer           regression    linear_reg   <spec[+]> 
 6         6 glmnet          regression    linear_reg   <spec[+]> 
 7         7 gls             regression    linear_reg   <spec[+]> 
 8         8 lme             regression    linear_reg   <spec[+]> 
 9         9 lmer            regression    linear_reg   <spec[+]> 
10        10 stan            regression    linear_reg   <spec[+]> 
# … with 36 more rows

So we see we get a nicely generated tibble of output that matchs a model spec to the .model_id and to the appropriate parsnip engine and mode

We can also choose the models we may want by giving either arguments to the .parsnip_engine parameter or .parsnip_fns or both.

library(dplyr)

fast_regression_parsnip_spec_tbl(.parsnip_fns = "linear_reg")
# A tibble: 11 × 5
   .model_id .parsnip_engine .parsnip_mode .parsnip_fns model_spec
       <int> <chr>           <chr>         <chr>        <list>    
 1         1 lm              regression    linear_reg   <spec[+]> 
 2         2 brulee          regression    linear_reg   <spec[+]> 
 3         3 gee             regression    linear_reg   <spec[+]> 
 4         4 glm             regression    linear_reg   <spec[+]> 
 5         5 glmer           regression    linear_reg   <spec[+]> 
 6         6 glmnet          regression    linear_reg   <spec[+]> 
 7         7 gls             regression    linear_reg   <spec[+]> 
 8         8 lme             regression    linear_reg   <spec[+]> 
 9         9 lmer            regression    linear_reg   <spec[+]> 
10        10 stan            regression    linear_reg   <spec[+]> 
11        11 stan_glmer      regression    linear_reg   <spec[+]> 
fast_regression_parsnip_spec_tbl(.parsnip_eng = c("lm","glm"))
# A tibble: 3 × 5
  .model_id .parsnip_engine .parsnip_mode .parsnip_fns model_spec
      <int> <chr>           <chr>         <chr>        <list>    
1         1 lm              regression    linear_reg   <spec[+]> 
2         2 glm             regression    linear_reg   <spec[+]> 
3         3 glm             regression    poisson_reg  <spec[+]> 
fast_regression_parsnip_spec_tbl(.parsnip_eng = "glm") %>%
  pull(model_spec)
[[1]]
Linear Regression Model Specification (regression)

Computational engine: glm 


[[2]]
Poisson Regression Model Specification (regression)

Computational engine: glm 

Voila!

To leave a comment for the author, please follow the link and comment on their blog: Steve's Data Tips and Tricks.

R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)