SPM Contribution Diagnostics

When an alarm is triggered, identifies which functional variables or components contribute most to the elevated T-squared or SPE statistic. Useful for root-cause analysis.

Usage

spm.contributions(
  scores,
  eigenvalues = NULL,
  grid.sizes = NULL,
  type = c("t2", "spe"),
  standardized.vars = NULL,
  reconstructed.vars = NULL,
  argvals.list = NULL
)

Arguments

scores

Score matrix (n x ncomp).

eigenvalues

Eigenvalues vector (length ncomp). Required for T-squared.

grid.sizes

Integer vector of grid sizes per variable. For univariate SPM, use a single value equal to ncomp.

type

Character; either "t2" for T-squared contributions or "spe" for SPE contributions. Default "t2".

standardized.vars

For SPE contributions: list of standardized data matrices.

reconstructed.vars

For SPE contributions: list of reconstructed data matrices.

argvals.list

For SPE contributions: list of argvals vectors.

Value

A matrix (n x p) of per-variable contributions.

Examples

# \donttest{
set.seed(1)
scores <- matrix(rnorm(10 * 3), 10, 3)
eigenvalues <- c(5, 2, 1)
# Per-component contributions (univariate, 1 variable with 3 grid points)
contrib <- spm.contributions(scores, eigenvalues, grid.sizes = 3L)
dim(contrib)
#> [1] 10  1
# }