Struct WnetResult 

#[non_exhaustive]
pub struct WnetResult {
    pub intercept: f64,
    pub beta_t: Vec<f64>,
    pub fitted_values: Vec<f64>,
    pub residuals: Vec<f64>,
    pub coeff_weights: Vec<f64>,
    pub selected: Vec<usize>,
    pub lambda: f64,
    pub alpha: f64,
    pub family: WaveletFamily,
    pub mode: BoundaryMode,
    pub level: usize,
}

Result of a wnet fit.

Carries the time-domain functional coefficient β(t), the sparse coefficient- space weights it was reconstructed from, the indices of the nonzero (selected) coefficients, the CV-selected λ, the elastic-net mixing alpha, fitted values / residuals, and the DWT configuration a future predict (Phase 71) needs to reproduce the transform.

Fields (Non-exhaustive)

Non-exhaustive structs could have additional fields added in future. Therefore, non-exhaustive structs cannot be constructed in external crates using the traditional Struct { .. } syntax; cannot be matched against without a wildcard ..; and struct update syntax will not work.

intercept: f64

Affine intercept α such that ŷ_i = intercept + Σ_j design[i,j] · coeff_weights[j] reproduces the fitted values directly (the elastic-net affine intercept, matching the predict formula).

beta_t: Vec<f64>

Time-domain functional coefficient β(t) (length m = curve length).

fitted_values: Vec<f64>

Fitted response values (length n).

residuals: Vec<f64>

Residuals y - ŷ (length n).

coeff_weights: Vec<f64>

Coefficient-space functional coefficient (length P = total wavelet coefficients) — sparse (many exact zeros).

selected: Vec<usize>

Indices (into coeff_weights) of the nonzero/selected coefficients.

lambda: f64

Cross-validation-selected λ.

alpha: f64

Elastic-net mixing parameter used (config.alpha).

family: WaveletFamily

Wavelet family used for the DWT (for reproducing the transform in prediction).

mode: BoundaryMode

Boundary mode used for the DWT.

level: usize

Effective decomposition depth used.

Implementations

impl WnetResult

pub fn predict(&self, new: &FdMatrix) -> Result<Vec<f64>, FdarError>

Predict the scalar response for new functional curves (WAV-05).

Re-transforms each new curve into the wavelet-coefficient design using the STORED fitted DWT configuration (family / mode / effective level), then applies the affine coefficient-space map ŷ = intercept + Σ_j design[i,j] · coeff_weights[j]. Re-passing the training curves reproduces the stored fitted_values exactly (up to float rounding).

Arguments

• new — functional predictor matrix (rows = curves) on the SAME evaluation grid as the training data (new.ncols() must equal the training grid length).

Errors

• FdarError::InvalidDimension with parameter: "new" if new has zero rows (no curves to predict), if new.ncols() differs from the training grid length, or (defensively) if the re-transformed design width disagrees with the stored coefficient-space width.
• FdarError::InvalidParameter if the DWT rejects the stored family/level (surfaced from decompose_matrix).

pub fn beta_t(&self) -> &[f64]

The time-domain functional coefficient β(t) (length m = curve length).

pub fn coefficient_function(&self) -> &[f64]

The functional coefficient β(t) (crate-convention alias of beta_t).

pub fn fitted_values(&self) -> &[f64]

The fitted response values (length n).

Trait Implementations

impl Clone for WnetResult

fn clone(&self) -> WnetResult

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for WnetResult

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

Formats the value using the given formatter.

impl PartialEq for WnetResult

fn eq(&self, other: &WnetResult) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for WnetResult