Per-PC Contributions to T-Squared Statistic

Computes the contribution of each principal component to the Hotelling T-squared statistic for each observation. Useful for diagnosing which components drive an alarm.

Usage

spm.pc.contributions(scores, eigenvalues)

Arguments

scores

Score matrix (n x ncomp), e.g., from spm.monitor()$scores.

eigenvalues

Numeric vector of eigenvalues (length ncomp).

Value

A matrix (n x ncomp) of per-component T-squared contributions. Row sums equal the T-squared statistics.

See also

spm.contributions for per-variable contributions

Examples

# \donttest{
set.seed(1)
scores <- matrix(rnorm(10 * 3), 10, 3)
eigenvalues <- c(5, 2, 1)
contrib <- spm.pc.contributions(scores, eigenvalues)
dim(contrib)  # 10 x 3
#> [1] 10  3
# }