Skip to main content

two_sample_mean_test

fdars_core::inference

Function two_sample_mean_test 

Source 
pub fn two_sample_mean_test(
    data_a: &FdMatrix,
    data_b: &FdMatrix,
    argvals: &[f64],
    ncomp: usize,
) -> Result<TestResult, FdarError>
Expand description

Functional two-sample mean-equality test via Hotelling-T² on a shared FPC basis (fda.usc-style mean equality).

Both samples are projected onto a common FPC basis fitted on the pooled data. The Hotelling-T² statistic is formed on the difference of the two group score-means, scaled by the effective sample size sqrt(n_a · n_b / (n_a + n_b)) so that under the null the statistic is asymptotically χ²(ncomp). The p-value is the χ²(ncomp) upper-tail probability of the observed statistic.

The eigenvalues fed to hotelling_t2 are derived from the pooled FPCA singular values via eigenvalue = sv² / (n_pooled − 1) (the mfpca convention).

§Arguments

  • data_a - First sample (n_a x m).
  • data_b - Second sample (n_b x m).
  • argvals - Evaluation points (length m).
  • ncomp - Number of FPC components for the shared basis.

Returns a TestResult with n_perm = 0 (non-permutation path).

§Errors

Returns FdarError::InvalidDimension if the two samples have unequal or zero column counts, if argvals.len() does not match the column count, or if either sample has fewer than 2 rows. Returns FdarError::InvalidParameter if ncomp < 1. Propagates errors from fdata_to_pc_1d / hotelling_t2.