Jackknife+ Regression — jackknife.plus • fdars

Constructs prediction intervals using jackknife+ (leave-one-out conformal). Most sample-efficient but most expensive method (n refits).

Usage

jackknife.plus(
  fdataobj,
  y,
  newdata,
  method = c("fregre.lm", "fregre.np"),
  scalar.train = NULL,
  scalar.test = NULL,
  ncomp = 3,
  h.func = 0,
  h.scalar = 0,
  alpha = 0.1,
  seed = NULL
)

Arguments

fdataobj: An object of class 'fdata' (training data).
y: Response vector (training).
newdata: An object of class 'fdata' (test data).
method: Regression method: "fregre.lm" or "fregre.np".
scalar.train: Optional scalar covariates for training.
scalar.test: Optional scalar covariates for test.
ncomp: Number of FPC components (default 3, for fregre.lm).
h.func: Functional bandwidth (default 0 = auto, for fregre.np).
h.scalar: Scalar bandwidth (default 0 = auto, for fregre.np).
alpha: Miscoverage level (default 0.1 for 90 percent intervals).
seed: Random seed (unused, kept for API consistency).

Value

Same as cv.conformal.regression.

Examples

# \donttest{
fd <- fdata(matrix(rnorm(500), 50, 10), argvals = seq(0, 1, length.out = 10))
y <- rnorm(50)
jp <- jackknife.plus(fd[1:40, ], y[1:40], fd[41:50, ])
# }