cv_fit0: Run initial cross validation fits for JADE

In jean997/jadeTF:

Description

Use an existing JADE fit at γ = 0 to start cross validation fits with the same parameters. These can be used as start computation of a series of solutions for γ between 0 and γ_max using either jade_path_plan or jade_path.

Usage

cv_fit0(orig.fit, n.folds = 5, which.fold = 1:n.folds, data.file = NULL,
  n.rep.bs.var = NULL, save.prefix = NULL, return.objects = TRUE,
  lambda = NULL)

Arguments

orig.fit	JADE fit using all the data or the name of a file containig a JADE fit.
n.folds	Number of cross validation folds.
which.fold	Vector of folds to fit. If This can be used to run cv fits in parallel. If which.fold=1:n.folds or is left missing cv fits will be run sequentially.
data.file	Optional file containing data. This is required if you want to bootstrap the variance.
n.rep.bs.var	Number of reps to use in bootstrapping variance of the fit. If NULL varaince will not be estimated.
save.prefix	If provided each fit will be saved to file named save.prefix.fold.RData
return.objects	If TRUE a list of jade objects will be returned. Otherwise nothing will be returned.
lambda	If NULL λ will be chosen by cross validation for each fold. Otherwise lambda should be a list of length which.fold giving lambda for each fold.

Value

A single jade object or list of jade objects of the same type as orig.fit (either jade_admm or jade_gd) if return.objects=TRUE

jean997/jadeTF documentation built on May 18, 2019, 11:44 p.m.