Holdout Resampling

Source: R/mlr_resamplings_holdout.R

Simple holdout: A single split into training and test. Parameter ratio determines the ratio of observation in the train set.

ResamplingHoldout

Format

R6::R6Class inheriting from Resampling.

Examples

# Create a task with 10 observations
task = mlr_tasks$get("iris")
task$filter(1:10)

# Instantiate Resampling
rho = mlr_resamplings$get("holdout")
rho$instantiate(task)

# Individual sets:
rho$train_set(1)
#> [1] 8 6 4 2 1 3 7
rho$test_set(1)
#> [1]  5  9 10
intersect(rho$train_set(1), rho$test_set(1))
#> integer(0)

# Internal storage:
rho$instance # simple list
#> $train
#> [1] 8 6 4 2 1 3 7
#> 
#> $test
#> [1]  5  9 10
#>