hublasso: Lasso with Huber's loss function

In Mufabo/Rrobustsp: Robust Signal Processing

Lasso with Huber's loss function

Description

hublasso computes the M-Lasso estimate for a given penalty parameter using Huber's loss function

Usage

hublasso(y, X, c = NULL, lambda, b0, sig0, reltol = 1e-05,
  printitn = 0, iter_max = 500)

Arguments

y:	Numeric data vector of size N x 1 (output,respones)
X:	Numeric data matrix of size N x p (inputs,predictors,features). Each row represents one observation, and each column represents one predictor
lambda:	positive penalty parameter value
b0:	numeric initial start of the regression vector
sig0:	numeric positive scalar, initial scale estimate.
c:	Threshold constant of Huber's loss function
reltol:	Convergence threshold. Terminate when successive estimates differ in L2 norm by a rel. amount less than reltol. Default is 1.0e-5
iter_max:	int, default = 500. maximum number of iterations
printitn:	print iteration number (default = 0, no printing)

Value

b0: regression coefficient vector estimate

sig0: estimate of the scale

psires: pseudoresiduals

Note

File in Regression.R

Examples

hublasso(rnorm(5), matrix(rnorm(5)), lambda = 0.5, b0 = rnorm(5), sig0 = 0.3)

Mufabo/Rrobustsp documentation built on June 11, 2022, 10:41 p.m.