Select Number of Principal Components

Selects the number of principal components to retain based on eigenvalues using one of several criteria: cumulative variance threshold, Kaiser rule, elbow method, or a fixed number.

Usage

spm.ncomp.select(
  eigenvalues,
  method = c("variance90", "kaiser", "elbow", "fixed"),
  threshold = 0.9
)

Arguments

eigenvalues

Numeric vector of eigenvalues from FPCA.

method

Selection method. One of "variance90" (cumulative variance exceeds threshold), "kaiser" (eigenvalues > mean), "elbow" (scree plot elbow), or "fixed" (use threshold as the number of components).

threshold

For "variance90", the cumulative proportion of variance to exceed (default 0.9). For "fixed", the exact number of components.

Value

An integer: the selected number of components.

Examples

# \donttest{
eigenvalues <- c(5.0, 2.0, 1.0, 0.5, 0.2, 0.1)
spm.ncomp.select(eigenvalues, method = "variance90", threshold = 0.9)
#> [1] 3
spm.ncomp.select(eigenvalues, method = "kaiser")
#> [1] 2
# }