M-Out-of-n Bootstrap Functions.

moonboot

This is an R package for m-out-of-n bootstrap that provides functions for

• computation of confidence intervals
• estimation of the scaling factor tau
• different methods for estimating m

Usage

A typical usage is

# application of estimator to the subset indices
boot.stat <- function(dat, indices) {
  my.stat(dat[indices])
}

# apply m-out-of-n bootstrap
boot.out <- mboot(x, boot.stat, m=2*sqrt(length(x)))

# compute 95% confidence interval
ci <- mboot.ci(boot.out, type="basic")
print(ci)

Beware that the last command estimates the scaling factor, which can be quite unreliable. It is thus better to provide the scaling factor if it is known, e.g., for a root-n consistent estimator:

ci <- mboot.ci(boot.out, tau=function(n) { n^0.5 }, type="basic")

Installation

If you have cloned the repository to the directory moonboot, you can install it with

R CMD INSTALL moonboot

If you have downloaded a file release moonboot-X.Y.Z.tar.gz, you can install it with

R CMD INSTALL moonboot-X.Y.Z.tar.gz

If you want to remove the already installed package, you can uninstall it with

R CMD REMOVE moonboot

Reference

For details about the implemented methods and a comparative evaluation, see

Dalitz C. and Lögler M. (2024): "moonboot: An R Package Implementing m-out-of-n Bootstrap Methods." https://doi.org/10.48550/arXiv.2412.05032

Authors & Copyright

• Christoph Dalitz, 2024 https://lionel.kr.hsnr.de/~dalitz/
• Felix Lögler, 2024

For licensing information, see the file LICENSE for details.