sensitivity.lasso: Perturbation of weights

In tathagatabasu/bootlasso: Boot-strap Method for LASSO

Description

Sensitivity analysis of predictors using perturbed weights

Usage

sensitivity.lasso(x, y, wt = NULL, ts = NULL, method = lasso_cd,
  k = 5, n_it = 10, df = NULL, nsim = 50, rel_err = 1)

Arguments

x	predictors
y	response
wt	weights for the coefficients of weighted LASSO. Defaults to NULL
ts	stepsize for proximal gradient and sub-gradient method (use opt_ts() to generate stepsize). Defaults to NULL
method	lasso optimization function. Three different methods are available to use. method = c(lasso_cd, lasso_sg, lasso_pg). Defaults to lasso_cd
k	number of fold. Default value is 5.
n_it	number of iteration for lasso_cd method. Default value is 10.
df	Degree of freedom. Number of desired variables to be zero. Defaults to NULL
nsim	number of perturbed weights. Default value is 50
rel_err	Relative purturbed error. Default value is 1

Value

The function returns the summary of sensitivity analysis.

tathagatabasu/bootlasso documentation built on Aug. 9, 2019, 1:07 a.m.