Struct KernelPCA 

pub struct KernelPCA<F> { /* private fields */ }

Kernel PCA configuration.

Holds hyperparameters for the kernel PCA decomposition. Calling Fit::fit computes the kernel eigendecomposition and returns a FittedKernelPCA that can project new data via Transform::transform.

Implementations

impl<F: Float + Send + Sync + 'static> KernelPCA<F>

pub fn new(n_components: usize) -> Self

Create a new KernelPCA that retains n_components components.

Defaults: kernel=Linear, gamma=None (auto: 1/n_features), degree=3, coef0=0.

pub fn with_kernel(self, kernel: Kernel) -> Self

Set the kernel function.

pub fn with_gamma(self, gamma: f64) -> Self

Set the gamma parameter for RBF, polynomial, and sigmoid kernels.

pub fn with_degree(self, degree: usize) -> Self

Set the degree for the polynomial kernel.

pub fn with_coef0(self, coef0: f64) -> Self

Set the independent term for polynomial and sigmoid kernels.

pub fn n_components(&self) -> usize

Return the configured number of components.

pub fn kernel(&self) -> Kernel

Return the configured kernel.

pub fn gamma(&self) -> Option<f64>

Return the configured gamma, if any.

pub fn degree(&self) -> usize

Return the configured polynomial degree.

pub fn coef0(&self) -> f64

Return the configured coef0.

Trait Implementations

impl<F: Clone> Clone for KernelPCA<F>

fn clone(&self) -> KernelPCA<F>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<F: Debug> Debug for KernelPCA<F>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<F: Float + Send + Sync + 'static> Fit<ArrayBase<OwnedRepr<F>, Dim<[usize; 2]>>, ()> for KernelPCA<F>

fn fit(&self, x: &Array2<F>, _y: &()) -> Result<FittedKernelPCA<F>, FerroError>

Fit Kernel PCA by computing the kernel matrix, centring it in feature space, and eigendecomposing.

Errors

• FerroError::InvalidParameter if n_components is zero or exceeds the number of samples.
• FerroError::InsufficientSamples if there are fewer than 2 samples.
• FerroError::ConvergenceFailure if the Jacobi eigendecomposition does not converge.

type Fitted = FittedKernelPCA<F>

The fitted model type returned by fit.

type Error = FerroError

The error type returned by fit.

impl<F: Float + Send + Sync + 'static> PipelineTransformer<F> for KernelPCA<F>

fn fit_pipeline( &self, x: &Array2<F>, _y: &Array1<F>, ) -> Result<Box<dyn FittedPipelineTransformer<F>>, FerroError>

Fit KernelPCA using the pipeline interface.

The y argument is ignored; Kernel PCA is unsupervised.

Errors

Propagates errors from Fit::fit.