group.lasso.eval.parameters: Group lasso cross validation

In ManuSetty/SeqGL: Group lasso for Dnase/ChIP-seq data

Description

Using the spams toolbox to run group lasso at various parameters to identify parameter combination with best test auc.

Usage

group.lasso.eval.parameters(train.features, train.labels, test.features,
  test.labels, groups, lambdas = c(0.01, 0.005, 0.001, 5e-04, 1e-04),
  no.cores = 1)

Arguments

train.features	Training feature matrix
train.labels	Training labels (Should be +1/-1)
test.features	Test feature matrix
test.labels	Test labels (Should be +1/-1)
groups	Feature assignment to groups. Each feature should belong to exactly one group
lambdas	Vector of regularization parameters
no.cores	Number of cores for parallel processing

Value

List containing lambdas and a matrix of test aucs named auc.matrix

See Also

run.group.lasso

ManuSetty/SeqGL documentation built on March 14, 2021, 3:18 a.m.