Survey-weighted Functional on Scalar Regression

A high-level wrapper for running survey-weighted function-on-scalar regression with bootstrap inference.

Usage

svyfui(
  formula,
  data,
  weights = NULL,
  family = gaussian(),
  boot_type = "weighted",
  num_boots = 500,
  nknots_min = NULL,
  nknots_min_fpca = NULL,
  seed = 2025,
  conf_level_pw = 0.95,
  conf_level_joint = 0.95,
  verbose = TRUE,
  parallel = FALSE,
  n_cores = NULL,
  ...
)

Arguments

formula: Formula with functional outcome on predictors, e.g. Y ~ X1 + X2.
data: A data frame with functional outcome columns, predictors, weights, etc.
weights: Optional bare column name for weights or external weight vector
family: Outcome distribution family (e.g., "gaussian", "binomial").
boot_type: Bootstrap method: "BRR", "Rao-Wu-Yue-Beaumont", "weighted", "unweighted".
num_boots: Number of bootstrap replicates.
nknots_min: Minimum number of knots for smoothing (optional).
nknots_min_fpca: Minimum number of knots for FPCA (optional).
seed: Random seed for reproducibility.
conf_level_pw: Confidence level for pointwise confidence intervals (default 0.95).
conf_level_joint: Confidence level for joint confidence intervals (default 0.95).
verbose: Whether to print messages about progress
parallel: Whether to run bootstrap in parallel (default FALSE).
n_cores: Number of cores to use if parallel = TRUE
...: Additional arguments passed to helpers.

Value

A list with components:

betaHat: Smoothed coefficient functions
cis: Bootstrap confidence intervals
boots: Raw bootstrap draws of coefficients
tidy_df: Tidy data frame for plotting with columns l (functional domain), beta_hat (estimate), lower_pw (pointwise lower CI), upper_pw (pointwise upper CI), lower_joint (joint lower CI), upper_joint (joint upper CI), and var_name (variable name from regression)

Examples

# \donttest{
   fit <- svyfui(Y ~ X, data = svyfosr::sample_df, weights = weight)
#> Estimating coefficients
#> Smoothing coefficients
#> Bootstrapping
#> Obtaining pointwise and joint confidence intervals
#> Completed!
   plot(fit)

# }