modernBoot 0.1.0
New Features
- Initial release of modernBoot package with core
resampling methods
- Bootstrap methods:
bs_mean(),
bca_ci(), studentized_ci() for confidence
intervals
- Wild bootstrap:
wild_boot_lm() for
heteroscedastic linear models with Rademacher and Mammen weights
- Block bootstrap:
moving_block_boot()
and stationary_boot() for dependent time series data
- Permutation tests:
perm_test_2sample()
for two-sample inference
- Multiple testing correction:
perm_maxT() for controlling family-wise error rate
- Automated method selection:
auto_select_method() intelligently recommends resampling
approach based on data structure
- Simulation tools:
compare_methods_sim() for benchmarking different resampling
methods
- Parallel computation: Full support for
future package parallelization (user-controlled)
Functions
Bootstrap Functions
bs_mean() — Nonparametric bootstrap confidence interval
for the mean (percentile method)bca_ci() — Bias-corrected and accelerated (BCa)
bootstrap confidence intervalstudentized_ci() — Studentized bootstrap confidence
interval for quantiles
Dependent Data Bootstrap
moving_block_boot() — Moving block bootstrap for time
seriesstationary_boot() — Stationary bootstrap (Politis &
Romano, 1994)
Model-Based Bootstrap
wild_boot_lm() — Wild bootstrap for linear regression
with heteroscedasticity
Permutation Tests
perm_test_2sample() — Two-sample permutation testperm_maxT() — Permutation maxT for multiple hypothesis
testing with FWER control
Utilities
auto_select_method() — Automatic resampling method
selectioncompare_methods_sim() — Simulation comparison of
bootstrap methods- Internal helpers:
check_numeric_vector(),
.safe_sample()
Documentation
- Comprehensive package vignette:
vignette("method-selection")
- Full function documentation with examples
- README with quick-start guide and installation instructions
- CI/CD workflows (R-CMD-check on macOS, Linux, Windows; pkgdown
deployment)
Dependencies
- Imports:
stats, utils,
boot, future, future.apply
- Suggests:
testthat, covr,
pkgdown, knitr, rmarkdown,
rhub
- Requires: R ≥ 4.0
Testing
- Initial test suite included in
tests/testthat/
- Tests for
bs_mean() and
perm_test_2sample()
- Comprehensive test expansion planned for future releases
References
- Efron, B., & Tibshirani, R. J. (1993). An introduction to
the bootstrap. Chapman and Hall/CRC.
- Politis, D. N., & Romano, J. P. (1994). The stationary
bootstrap. Journal of the American Statistical Association,
89(428), 1303-1313.
- Wu, C. F. (1986). Jackknife, bootstrap and other resampling methods
in regression analysis. Annals of Statistics, 14(4),
1261-1295.
Future Enhancements
- Cluster bootstrap for clustered data
- Bayesian bootstrap methods
- Rcpp implementations for speed optimization
- Extended vignettes with real-world examples
- Performance benchmarking suite