Daim (version 1.1.0)

Daim.control: Control parameters for the diagnostic accuracy of models.

Description

Control of resampling methods.

Usage

Daim.control(method="boot", number = 100, replace = TRUE,  boot.size = 1, k = 10, k.runs = 1, dependency = list(var = NULL, keep.id = FALSE))

Arguments

method

the resampling method: boot - bootstrapping, cv - k-fold cross-validation.

number

the number of bootstrap samples

replace

a logical indicating whether sampling of observations is done with or without replacement

boot.size

percentage of observations (0 < boot.size < 1) to draw without replacement (only relevant if replace = FALSE). In this case subsampling is performed.

k

the number of folds

k.runs

the number of runs of k-fold cross-validations

dependency

to handle paired data structures

References

Breiman L. (1996), Out-Of-Bag Estimation. Technical Report. ftp://ftp.stat.berkeley.edu/pub/users/breiman/OOBestimation.ps.Z.

Adler W, Brenning A, Potapov S, Schmid M, Lausen B. (2011), Ensemble classification of paired data. Computational Statistics and Data Analysis. 55(5):1933-1941.

See Also

Daim

Examples

  ###
  ### for bootstrap
  ###

  Daim.control(method="boot", number=100)

  ###
  ### for cross-validation
  ###

  Daim.control(method="cv", k=10, k.runs=10)

  ###
  ### for subagging or subsampling
  ###

  Daim.control(method="boot", number=100, replace=FALSE, boot.size=0.9)