Skip to content

nullmodel models

Source: vignettes/nullmodel.Rmd
nullmodel.Rmd
Function Works
tidypredict_fit(), tidypredict_sql(), parse_model() ✔
tidypredict_to_column() ✔
tidypredict_test() ✔
tidypredict_interval(), tidypredict_sql_interval() ✗
parsnip ✔

parsnip::nullmodel() fits a model that ignores the predictors entirely. For regression it predicts the mean of the outcome, and for classification it predicts the observed class frequencies. Predictions are therefore constants, and tidypredict_fit() returns a plain number for regression and a named list of one constant per class for classification. Since the classification output is a list, tidypredict_to_column() and tidypredict_test() are only supported for regression.

tidypredict_ functions 

model <- parsnip::nullmodel(mtcars[-1], mtcars$mpg)

  • Create the R formula

    tidypredict_fit(model)
#> [1] 20.09062

  • Add the predictions to the original table

    mtcars %>%
  tidypredict_to_column(model) %>%
  glimpse()
#> Rows: 32
#> Columns: 12
#> $ mpg  <dbl> 21.0, 21.0, 22.8, 21.4, 18.7, 18.1, 14.3, 24.4, 22.8, 19…
#> $ cyl  <dbl> 6, 6, 4, 6, 8, 6, 8, 4, 4, 6, 6, 8, 8, 8, 8, 8, 8, 4, 4,…
#> $ disp <dbl> 160.0, 160.0, 108.0, 258.0, 360.0, 225.0, 360.0, 146.7, …
#> $ hp   <dbl> 110, 110, 93, 110, 175, 105, 245, 62, 95, 123, 123, 180,…
#> $ drat <dbl> 3.90, 3.90, 3.85, 3.08, 3.15, 2.76, 3.21, 3.69, 3.92, 3.…
#> $ wt   <dbl> 2.620, 2.875, 2.320, 3.215, 3.440, 3.460, 3.570, 3.190, …
#> $ qsec <dbl> 16.46, 17.02, 18.61, 19.44, 17.02, 20.22, 15.84, 20.00, …
#> $ vs   <dbl> 0, 0, 1, 1, 0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1,…
#> $ am   <dbl> 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1,…
#> $ gear <dbl> 4, 4, 4, 3, 3, 3, 3, 4, 4, 4, 4, 3, 3, 3, 3, 3, 3, 4, 4,…
#> $ carb <dbl> 4, 4, 1, 1, 2, 1, 4, 2, 2, 4, 4, 3, 3, 3, 4, 4, 4, 1, 2,…
#> $ fit  <dbl> 20.09062, 20.09062, 20.09062, 20.09062, 20.09062, 20.090…

  • Confirm that the results match the model’s predict() results

    tidypredict_test(model, mtcars)
#> tidypredict test results
#> Difference threshold: 1e-12
#> 
#>  All results are within the difference threshold

For classification, one expression per class is returned:

c_model <- parsnip::nullmodel(iris[-5], iris$Species)

tidypredict_fit(c_model)
#> $setosa
#> [1] 0.3333333
#> 
#> $versicolor
#> [1] 0.3333333
#> 
#> $virginica
#> [1] 0.3333333

parsnip

parsnip fitted models are also supported by tidypredict:

library(parsnip)

p_model <- null_model(mode = "regression") %>%
  set_engine("parsnip") %>%
  fit(mpg ~ ., data = mtcars)
tidypredict_fit(p_model)
#> [1] 20.09062

Parse model spec

Here is an example of the model spec:

pm <- parse_model(model)
str(pm, 2)
#> List of 2
#>  $ general:List of 4
#>   ..$ model  : chr "nullmodel"
#>   ..$ version: num 2
#>   ..$ type   : chr "regression"
#>   ..$ is_glm : num 0
#>  $ terms  :List of 1
#>   ..$ :List of 4
#>  - attr(*, "class")= chr [1:3] "parsed_model" "pm_regression" "list"