robregcc_option: Control parameter for model estimation:

In robregcc: Robust Regression with Compositional Covariates

Description

The model approach use scaled lasoo approach for model selection.

Usage

robregcc_option(
  maxiter = 10000,
  tol = 1e-10,
  nlam = 100,
  out.tol = 1e-08,
  lminfac = 1e-08,
  lmaxfac = 10,
  mu = 1,
  nu = 1.05,
  sp = 0.3,
  gamma = 2,
  outMiter = 3000,
  inMiter = 500,
  kmaxS = 500,
  tolS = 1e-04,
  nlamx = 20,
  nlamy = 20,
  spb = 0.3,
  spy = 0.3,
  lminfacX = 1e-06,
  lminfacY = 0.01,
  kfold = 10,
  fullpath = 0,
  sigmafac = 2
)

Arguments

maxiter	maximum number of iteration for convergence
tol	tolerance value set for convergence
nlam	number of lambda to be genrated to obtain solution path
out.tol	tolernce value set for convergence of outer loop
lminfac	a multiplier of determing lambda_min as a fraction of lambda_max
lmaxfac	a multiplier of lambda_max
mu	penalty parameter used in enforcing orthogonality
nu	penalty parameter used in enforcing orthogonality (incremental rate of mu)
sp	maximum proportion of nonzero elements in shift parameter
gamma	adaptive penalty weight exponential factor
outMiter	maximum number of outer loop iteration
inMiter	maximum number of inner loop iteration
kmaxS	maximum number of iteration for fast S estimator for convergence
tolS	tolerance value set for convergence in case of fast S estimator
nlamx	number of x lambda
nlamy	number of y lambda
spb	sparsity in beta
spy	sparsity in shift gamma
lminfacX	a multiplier of determing lambda_min as a fraction of lambda_max for sparsity in X
lminfacY	a multiplier of determing lambda_min as a fraction of lambda_max for sparsity in shift parameter
kfold	nummber of folds for crossvalidation
fullpath	1/0 to get full path yes/no
sigmafac	multiplying factor for the range of standard deviation

Value

a list of controling parameter.

Examples

 
# default options
library(robregcc)
control_default = robregcc_option()
# manual options
control_manual <- robregcc_option(maxiter=1000,tol = 1e-4,lminfac = 1e-7)

robregcc documentation built on July 26, 2020, 1:07 a.m.