Apply compositional Mahalanobis distance to new samples

Projects new samples into the same log-ratio space and computes their squared Mahalanobis distance using the frozen center and inverse-covariance from fit_compositional_mahalanobis.

Usage

apply_compositional_mahalanobis(fit, x_test, return_pvalue = TRUE)

Arguments

fit

A compositional_mahalanobis_fit object.

x_test

Numeric matrix (samples \(\times\) features). All values must be non-negative. All training features must be present.

return_pvalue

Logical. If TRUE (default), also returns chi-square reference p-values.

Value

List with distance_sq, distance, and (if return_pvalue) pvalue_chisq.