control_params: Create control parameter object for the classifiers
In bootfs: Derive Robust Feature Sets for Two or Multiclass Classification Problems

Description Usage Arguments Details Value See Also Examples

This function creates a set of control parameters which is passed to the classifier functions.

  control_params(seed = 123, bstr = 100, ncv = 5,
    repeats = 10, saveres = TRUE, jitter = FALSE,
    maxiter = 1000, maxevals = 500, bounds = NULL,
    max_allowed_feat = NULL, n.threshold = 50,
    maxRuns = 300, localImp = TRUE,
    rfimportance = "MeanDecreaseAccuracy", ntree = 1000,
    shrinkage = 0.01, interaction.depth = 3,
    bag.fraction = 0.75, train.fraction = 0.75,
    n.minobsinnode = 3, n.cores = 1, verbose = TRUE)

`seed`	A random seed to be set before the classification
`bstr`	Integer. Number of bootstrap iterations.
`ncv`	Integer. Number of crossvalidation folds.
`repeats`	Integer. Number of repeats for cross-validation.
`saveres`	Boolean. If TRUE, save results.
`jitter`	Boolean. If TRUE, generate a small amount of noise, if standard deviations for samples are zero. NOTE: Use with care!
`maxiter`	Integer. Maximum number of iterations in SCAD SVM. Parameter for SCAD SVM from `penalizedSVM` package.
`maxevals`	Integer. Parameter for SCAD SVM from `penalizedSVM` package.
`bounds`	Parameter for SCAD SVM from `penalizedSVM` package.
`max_allowed_feat`	Integer. PAMR parameter, bounding the maximum number of features reported.
`n.threshold`	Integer. PAMR parameter, number of thresholds to be generated.
`maxRuns`	Integer. RF_Boruta parameter, number of runs in Boruta selection.
`localImp`	Boolean. randomForest parameter; save local importances.
`rfimportance`	String. randomForest parameter; which importance measure should be used in the randomForest (method 'rf') to rank and select features? Either `MeanDecreaseGini` or `MeanDecreaseAccuracy`. Features are selected with `rfimportance` >= 1.
`ntree`	Integer. randomForest and GBM parameter; Number of trees to be used.
`shrinkage`	Double. GBM parameter; shrinkage step size.
`interaction.depth`	Integer. GBM parameter.
`bag.fraction`	Numeric in 0..1. GBM parameter; Fraction of bagged samples.
`train.fraction`	Numeric in 0..1. GBM paramter; Fraction of training samples.
`n.minobsinnode`	Integer. GBM parameter.
`n.cores`	Integer. GBM parameter.
`verbose`	Boolean. GBM parameter. Be verbose or not.

This function is used to define a set of control parameters used in the different methods. For each parameter, consult the respective help pages of the methodologies.

List with all named control parameters.

penalizedSVM randomForest gbm Boruta pamr