fdars (version 0.3.3)

fdata.bootstrap: Bootstrap Functional Data

Description

Generate bootstrap samples from functional data. Supports naive bootstrap (resampling curves with replacement) and smooth bootstrap (adding noise based on estimated covariance structure).

Usage

fdata.bootstrap(
  fdataobj,
  n.boot = 200,
  method = c("naive", "smooth"),
  variance = NULL,
  seed = NULL
)

Value

A list of class 'fdata.bootstrap' with components:

boot.samples

List of n.boot fdata objects, each a bootstrap sample

original

The original fdata object

method

The bootstrap method used

n.boot

Number of bootstrap replications

Arguments

fdataobj

An object of class 'fdata'.

n.boot

Number of bootstrap replications (default 200).

method

Bootstrap method: "naive" for resampling with replacement, "smooth" for adding Gaussian noise (default "naive").

variance

For method="smooth", the variance of the added noise. If NULL, estimated from the data.

seed

Optional seed for reproducibility.

Examples

# Create functional data
t <- seq(0, 1, length.out = 50)
X <- matrix(0, 20, 50)
for (i in 1:20) X[i, ] <- sin(2*pi*t) + rnorm(50, sd = 0.1)
fd <- fdata(X, argvals = t)

# Naive bootstrap
boot_naive <- fdata.bootstrap(fd, n.boot = 100, method = "naive")

# Smooth bootstrap
boot_smooth <- fdata.bootstrap(fd, n.boot = 100, method = "smooth")