EAlasso (version 0.1.0)

PB.CI: Provide (1-alpha)% confidence interval of each coefficients

Description

Using samples drawn by PBsampler, computes (1-alpha)% confidence interval of each coefficient.

Usage

PB.CI(object, alpha = 0.05, method = "debias", parallel = FALSE,
  ncores = 2L)

Arguments

object

bootstrap samples of class PB from PBsampler

alpha

significance level.

method

bias-correction method. Either to be "none" or "debias".

parallel

logical. If TRUE, use parallelization. Default is FALSE.

ncores

integer. The number of cores to use for parallelization.

Value

(1-alpha)% confidence interval of each coefficients

Details

If method==none, PB.CI simply compute the two-soded (1-alpha) quantile of the sampled coefficients. If method==debias, we use debiased estimator to compute confidence interval.

References

Zhang, C., Zhang, S. (2014), "Confidence intervals for low dimensional parameters in high dimensional linear models," Journal of the Royal Statistical Society: Series B, 76, 217<U+2013>242.

Dezeure, R., Buehlmann, P., Meier, L. and Meinshausen, N. (2015), "High-Dimensional Inference: Confidence Intervals, p-values and R-Software hdi," Statistical Science, 30(4), 533-558

Examples

# NOT RUN {
set.seed(1234)
n <- 40
p <- 50
Niter <-  10
Group <- rep(1:(p/10), each = 10)
Weights <- rep(1, p/10)
X <- matrix(rnorm(n*p), n)
object <- PBsampler(X = X, PE_1 = c(1,1,rep(0,p-2)), sig2_1 = 1, lbd_1 = .5,
niter = 100, type = "lasso")
parallel <- (.Platform$OS.type != "windows")
PB.CI(object = object, alpha = .05, method = "none")

# }