cv.lambda: Perform cross-validation to select the best fit and finds the...

In Saptati-Datta/AdapLasso: Computes Adaptive Lasso Using Lars Algorithm

Description

Perform cross-validation to select the best fit and finds the optimal lambda for a particular gamma value

Usage

cv.lambda(
  X,
  Y,
  tuning_seq = NULL,
  len_tuning = 60,
  gamma = 0.01,
  k = 5,
  id_fold = NULL,
  eps = 0.001
)

Arguments

X	n x p design matrix of inputs
Y	n x 1 vector of outputs
tuning_seq	(optional)sequence of tuning parameters
len_tuning	length of desired tuning parameter sequence
gamma	a scalar(>0) input used in the weight(user input)
k	number of folds for k-fold cross-validation, default is 5
id_fold	(optional) vector of length n specifying the folds assignment (from 1 to max(folds_ids)), if supplied the value of k is ignored
eps	precision level for convergence assessment, default 0.001

Value

tuning_seq	the actual sequence of tuning parameters used
beta_lamb	p x length(tuning_seq) matrix of corresponding solutions at each lambda value (original data without center or scale)
intercept_vec	Unscaled vector of intercepts for a fixed gamma and for different lambda values
id_fold	used splitting into folds from 1 to k (either as supplied or as generated in the beginning)
lambda_min	selected lambda based on minimal rule
cv	values of CV(lambda) for each lambda

Examples

X <- matrix(rnorm(500), 50, 10)
Y <- rnorm(50)
fit_cv <- cv.lambda(X, Y)

Saptati-Datta/AdapLasso documentation built on Dec. 18, 2021, 12:57 p.m.