Skip to contents

Block Bootstrap with Diagnostics

blockBootstrap_with_diagnostics.Rd

Performs Moving Block Bootstrap or Stationary Bootstrap while collecting detailed diagnostic information about block composition.

Usage

blockBootstrap_with_diagnostics(
  x,
  n_boot = NULL,
  block_length = NULL,
  bs_type = c("moving", "stationary"),
  block_type = c("overlapping", "non-overlapping"),
  num_boots = 100L,
  p = 0.1,
  collect_diagnostics = TRUE,
  taper_type = "cosine",
  tukey_alpha = 0.5
)

Arguments

x

Numeric matrix with rows as time points, columns as variables.

n_boot

Integer. Desired length of bootstrap series. If NULL, uses length of original series.

block_length

Integer. Block length for moving bootstrap, or expected block length for stationary bootstrap. If NULL, computed automatically.

bs_type

Character. Either "moving" or "stationary".

block_type

Character. One of "overlapping", "non-overlapping", or "tapered".

num_boots

Integer. Number of bootstrap replicates.

p

Numeric in (0,1). Probability parameter for geometric distribution in stationary bootstrap. Default 0.1.

collect_diagnostics

Logical. If TRUE, collect block-level diagnostics.

taper_type

Character. Taper window type if block_type = "tapered".

tukey_alpha

Numeric. Alpha parameter for Tukey window.

Value

If collect_diagnostics = FALSE, a list of bootstrap matrices. If collect_diagnostics = TRUE, a list with:

bootstrap_series

List of bootstrap matrices

diagnostics

A tsbs_diagnostics object with block information

Details

This function provides an R implementation of block bootstrap that tracks diagnostic information. For each bootstrap replicate, it records:

  • Block lengths used

  • Starting positions in original series

  • Source indices mapping each bootstrap observation to original

For the Moving Block Bootstrap, blocks have fixed length and are sampled uniformly from all possible starting positions.

For the Stationary Bootstrap, block lengths are drawn from a geometric distribution with parameter p, giving expected block length 1/p.

Examples

if (FALSE) { # \dontrun{
set.seed(123)
x <- matrix(rnorm(500), ncol = 2)

# Moving block bootstrap with diagnostics
result <- blockBootstrap_with_diagnostics(
  x, block_length = 10, bs_type = "moving",
  num_boots = 50, collect_diagnostics = TRUE
)

# View block length distribution
summary(result$diagnostics)

# Plot coverage
plot_block_coverage(result$diagnostics)
} # }