Function pca

pub fn pca(
    data: &Tensor,
    n_components: usize,
) -> Result<(Tensor, Tensor, Vec<f64>), RuntimeError>

Expand description

Principal Component Analysis via SVD of centered data.

data is a 2D Tensor of shape (n_samples, n_features). n_components is the number of principal components to keep.

Returns (transformed_data, components, explained_variance_ratio):

transformed_data: (n_samples, n_components) — data projected onto principal components
components: (n_components, n_features) — principal component directions (rows)
explained_variance_ratio: Vec of length n_components — fraction of variance per component

Determinism contract: All reductions use BinnedAccumulatorF64.

pca