tune (version 0.1.2)

min_grid.model_spec: Determine the minimum set of model fits

Description

min_grid() determines exactly what models should be fit in order to evaluate the entire set of tuning parameter combinations. This is for internal use only and the API may change in the near future.

Usage

# S3 method for model_spec
min_grid(x, grid, ...)
fit_max_value(x, grid, ...)
# S3 method for boost_tree
min_grid(x, grid, ...)
# S3 method for linear_reg
min_grid(x, grid, ...)
# S3 method for logistic_reg
min_grid(x, grid, ...)
# S3 method for mars
min_grid(x, grid, ...)
# S3 method for multinom_reg
min_grid(x, grid, ...)
# S3 method for nearest_neighbor
min_grid(x, grid, ...)
# S3 method for cubist_rules
min_grid(x, grid, ...)
# S3 method for C5_rules
min_grid(x, grid, ...)
# S3 method for rule_fit
min_grid(x, grid, ...)
# S3 method for pls
min_grid(x, grid, ...)
# S3 method for poisson_reg
min_grid(x, grid, ...)

Arguments

x

A model specification.

grid

A tibble with tuning parameter combinations.

...

Not currently used.

Value

A tibble with the minimum tuning parameters to fit and an additional list column with the parameter combinations used for prediction.

Details

fit_max_value() can be used in other packages to implement a min_grid() method.

Examples

# NOT RUN {
library(dplyr)
library(dials)
library(parsnip)

## -----------------------------------------------------------------------------
## No ability to exploit submodels:

svm_spec <-
  svm_poly(cost = tune(), degree = tune()) %>%
  set_engine("kernlab") %>%
  set_mode("regression")

svm_grid <-
  svm_spec %>%
  parameters() %>%
  grid_regular(levels = 3)

min_grid(svm_spec, svm_grid)

## -----------------------------------------------------------------------------
## Can use submodels

xgb_spec <-
  boost_tree(trees = tune(), min_n = tune()) %>%
  set_engine("xgboost") %>%
  set_mode("regression")

xgb_grid <-
  xgb_spec %>%
  parameters() %>%
  grid_regular(levels = 3)

min_grid(xgb_spec, xgb_grid)

# }