Skip to content

Summarize a final fit

Source: R/nested-final-fit-print.R
summary.nested_final_fit.Rd

summary() returns the facts of a final fit as values: the tuning run on all the data, the tuner and its counts, how many candidates it scored, and the selected values.

Usage

# S3 method for class 'nested_final_fit'
summary(object, ...)

# S3 method for class 'summary.nested_final_fit'
print(x, ...)

Arguments

object

A nested_final_fit from nested_final_fit().

...

Not used, and must be empty.

x

A summary.nested_final_fit object from summary.nested_final_fit().

Value

summary() returns a summary.nested_final_fit, a list with these components:

  • tuning_label, the tuning run's resampling label

  • tuner, the tune or finetune function that ran

  • candidates, the number of candidates the run scored

  • initial, initial_requested, iterations_completed and iterations_requested, the counts of an iterating search

  • selection, the selected parameter values

  • select and first_metric, as extract_procedure() records them

  • estimate, always NULL

print() returns x, invisibly.

Details

The estimate component is always NULL, because the number to report is not stored on the fit. It is the nested estimate from collect_metrics() on the nested run the fit was built from. The nested estimate describes the procedure that produced this model, not the fitted model itself.

The four iteration counts are NULL for a grid search or a race. On a fit with nothing to tune, candidates is 0, selection is empty, and the other components but tuner are NULL.

See also

print.nested_final_fit(), nested_final_fit(), summary.nested_results(), extract_tune_results()

Examples

data(mtcars)

rec <- recipes::recipe(mpg ~ ., data = mtcars) |>
  recipes::step_pca(recipes::all_predictors(), num_comp = tune::tune())
wf <- workflows::workflow(rec, parsnip::linear_reg())

set.seed(1)
folds <- nested_resamples(
  mtcars,
  outside = rsample::vfold_cv(v = 2),
  inside = rsample::vfold_cv(v = 2)
)
set.seed(2)
res <- nested_tune_grid(wf, folds, grid = data.frame(num_comp = 1:2))
set.seed(3)
final <- nested_final_fit(wf, res)
summary(final)
#> 
#> ── Nested cross-validation final fit ──────────────────────────────────
#> Full-data tuning: 2-fold cross-validation
#> Procedure: grid search, 2 candidates scored
#> Candidates scored: 2
#> 
#> ── Selected parameters ──
#> 
#> Selecting metric: rmse
#> num_comp: 1
#> 
#> ── Estimate ──
#> 
#> ℹ Report the nested estimate from `collect_metrics()` on the results
#>   object this fit was built from. For a fit built from a workflow set,
#>   that is this workflow's rows of `collect_metrics()` on the set. That
#>   holds for a workflow chosen before seeing the set's estimates. It
#>   describes the procedure that produced this model, and it is the
#>   number to report for this model.
#> ℹ The tuning run above has metrics, but selection consumed them.
#>   `extract_tune_results()` reaches them, and every one is a
#>   selection-time quantity, optimistically biased as a claim about this
#>   model.