Multivariate Functional Principal Component Analysis

Extends univariate FPCA to handle multiple functional variables observed on potentially different grids. Variables are optionally weighted by their inverse standard deviation before joint SVD.

Usage

mfpca(fdataobj.list, ncomp = 5, weighted = TRUE)

Arguments

fdataobj.list: A list of fdata objects, one per functional variable. All must have the same number of observations (rows) but may differ in grid size.
ncomp: Number of principal components to extract (default 5).
weighted: Logical; whether to weight each variable by 1/std_dev before SVD (default TRUE).

Value

An object of class mfpca with components:

scores: Score matrix (n x ncomp)
eigenfunctions: List of eigenfunction matrices, one per variable
eigenvalues: Eigenvalues (length ncomp)
means: List of mean functions, one per variable
scales: Per-variable standard deviations
grid.sizes: Grid sizes per variable

Examples

# \donttest{
# Two functional variables
n <- 40
fd1 <- fdata(matrix(rnorm(n * 20), n, 20), argvals = seq(0, 1, length.out = 20))
fd2 <- fdata(matrix(rnorm(n * 15), n, 15), argvals = seq(0, 1, length.out = 15))
result <- mfpca(list(fd1, fd2), ncomp = 3)
dim(result$scores)
#> [1] 40  3
# }