Selection of the tuning parameters of desp by v-fold cross-validation

Description

This function returns the precision matrix and the expectation associated to the data matrix X using desp and choosing the tuning parameters lambda and gamma by v-fold cross-validation that uses a robust loss function. The expression of the loss function is provided in the companion vignette.

Usage

1
  desp.cv(X, v, lambda.range, gamma.range, settings=NULL)

Arguments

X

The data matrix.

v

The number of folds.

lambda.range

The range of the penalization parameter lambda that encourages robustness.

gamma.range

The range of the penalization parameter gamma that promotes sparsity.

settings

A list including all the parameters needed for the estimation. Please refer to the documentation of the function desp to get more details.

Value

desp.cv returns an object with S3 class "desp.cv" containing the estimated parameters along with the selected values of the tuning parameters, with components:

Omega

The precision matrix.

mu

The expectation vector.

Theta

The matrix corresponding to outliers.

lambda

The selected lambda.

gamma

The selected gamma.

Author(s)

Arnak Dalalyan and Samuel Balmand.

See Also

desp