crda.cv: Cross-validation based Joint-sparsity Level for CRDA Method

Description Usage Arguments Details Value Author(s) See Also Examples

Description

The function crda.cv performs cross-validation for finding the joint-sparsity level of CRDA method.

Usage

1
2
3
crda.cv(X, y, q = Inf, al = NULL, prior = "uniform", flds = NULL,
  nFolds = 5, Kgrid = round(exp(seq(log(0.05 * nrow(X)), log(0.5 *
  nrow(X)), length.out = 10))), centerX = FALSE, plotCV = FALSE)

Arguments

X

Training dataset, a pxn matrix with n-samples each having p-features.

y

Labels for training dataset, an nx1 vector of whole numbers.

q

Type of Lq,1 norm, default is Linf-norm.

al

Regularization parameter.

prior

Type of prior class probabilities, either 'uniform' (default) or 'estimated'.

flds

Folds for cross-validation (CV).

nFolds

number of folds.

Kgrid

A grid having candidate values of joint-sparsity level.

centerX

Flag for grand-mean centering of test dataset using grand-mean of training dataset.

plotCV

Flag for plotting the CV-results.

Details

crda.cv

Value

An object cv.obj of class crda.cv with the following attributes:

funCall

The call to the crda.cv function.

Kgrid

A grid having candidate values of joint-sparsity level.

folds

Folds for cross-validation (CV).

cvMER

CV misclassification error.

resubMER

Resubstitution error.

K

CV-estimate of joint-sparsity level.

regparam

The value of regularization parameter.

Author(s)

Muhammad Naveed Tabassum and Esa Ollila, 2018

See Also

crda.regparam, crda

Examples

1
2
3
4
crda.cv(X,y)
crda.cv(X,y, q = 1, nFolds = 10)
crda.cv(X,y, q = 1, prior = 'estimated')
crda.cv(X,y, q = 1, centerX = TRUE)

mntabassm/compressiveRDA documentation built on May 31, 2019, 5:22 p.m.