Rolling Window Cross Validation Resampling

Splits data using a folds-folds (default: 10 folds) rolling window cross-validation.

Format

R6::R6Class inheriting from Resampling.

Construction

ResamplingRollingWindowCV$new()
mlr3::mlr_resamplings$get("ResamplingRollingWindowCV")
rsmp("ResamplingRollingWindowCV")

Fields

See Resampling.

Methods

See Resampling.

Parameters

• folds :: integer(1)
  Number of folds.

• window_size :: integer(1)
  (Minimal) Size of the rolling window.

• horizon :: integer(1)
  Forecasting horizon in the test sets.

• fixed_window :: logial(1)
  Flag for fixed sized window. If FALSE an expanding window is used.

References

mlr3bischl_2012 paper:Ch. Bergmeir, R. J. Hyndman, B. Koo, A note on the validity of cross-validation for evaluating

See also

Dictionary of Resamplings: mlr_resamplings

as.data.table(mlr_resamplings) for a complete table of all (also dynamically created) Resampling implementations.

Examples

# Create a task with 10 observations
task = mlr3::tsk("airpassengers")
task$filter(1:20)

# Instantiate Resampling
rfho = mlr3::rsmp("RollingWindowCV", folds = 3, fixed_window = FALSE)
rfho$instantiate(task)

# Individual sets:
rfho$train_set(1)
#>  [1]  1  2  3  4  5  6  7  8  9 10
rfho$test_set(1)
#> [1] 11 12 13 14 15
intersect(rfho$train_set(1), rfho$test_set(1))
#> integer(0)

# Internal storage:
rfho$instance #  list
#> $train
#> $train[[1]]
#>  [1]  1  2  3  4  5  6  7  8  9 10
#> 
#> $train[[2]]
#>  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13
#> 
#> $train[[3]]
#>  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15
#> 
#> 
#> $test
#> $test[[1]]
#> [1] 11 12 13 14 15
#> 
#> $test[[2]]
#> [1] 14 15 16 17 18
#> 
#> $test[[3]]
#> [1] 16 17 18 19 20
#> 
#>