Struct MultiTaskElasticNetCV 

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

Multi-task ElasticNet regression with built-in cross-validation for joint (alpha, l1_ratio) selection.

For each candidate l1_ratio, generates a log-spaced alpha grid (from the L21 dual-norm alpha_max down to alpha_max * 1e-3), runs k-fold CV fitting a MultiTaskElasticNet per fold, and selects the combination minimizing the mean held-out MSE (averaged over samples and tasks).

Mirrors sklearn.linear_model.MultiTaskElasticNetCV (_coordinate_descent.py:2806).

Type Parameters

• F: The floating-point type (f32 or f64).

Implementations

impl<F: Float + FromPrimitive> MultiTaskElasticNetCV<F>

pub fn new() -> Self

Create a new MultiTaskElasticNetCV with default settings.

Defaults mirror sklearn.linear_model.MultiTaskElasticNetCV.__init__ (_coordinate_descent.py:2989-3018), which fixes a single l1_ratio=0.5:

• l1_ratios = [0.5]
• n_alphas = 100
• cv = 5
• max_iter = 1000
• tol = 1e-4
• fit_intercept = true

Use with_l1_ratios to search a grid of mixing ratios.

pub fn with_l1_ratios(self, l1_ratios: Vec<F>) -> Self

Set the candidate L1/L2 mixing ratios. Each value must be in [0, 1].

pub fn with_n_alphas(self, n_alphas: usize) -> Self

Set the number of alphas generated per l1_ratio.

pub fn with_cv(self, cv: usize) -> Self

Set the number of cross-validation folds (must be at least 2).

pub fn with_max_iter(self, max_iter: usize) -> Self

Set the maximum number of block-coordinate-descent iterations.

pub fn with_tol(self, tol: F) -> Self

Set the convergence tolerance.

pub fn with_fit_intercept(self, fit_intercept: bool) -> Self

Set whether to fit per-task intercept terms.

Trait Implementations

impl<F: Clone> Clone for MultiTaskElasticNetCV<F>

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

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<F: Debug> Debug for MultiTaskElasticNetCV<F>

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

Formats the value using the given formatter.

impl<F: Float + FromPrimitive> Default for MultiTaskElasticNetCV<F>

fn default() -> Self

Returns the “default value” for a type.

impl<F: Float + Send + Sync + ScalarOperand + FromPrimitive + 'static> Fit<ArrayBase<OwnedRepr<F>, Dim<[usize; 2]>>, ArrayBase<OwnedRepr<F>, Dim<[usize; 2]>>> for MultiTaskElasticNetCV<F>

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

Fit the MultiTaskElasticNetCV model.

For each candidate l1_ratio, generates the L21 alpha grid, runs k-fold CV for every (alpha, l1_ratio) pair (fitting a MultiTaskElasticNet per train fold and scoring held-out MSE), then refits on the full data using the best combination. Mirrors LinearModelCV.fit (_coordinate_descent.py:1552-1837).

Errors

• FerroError::ShapeMismatch if Y.nrows() differs from the number of samples in X.
• FerroError::InvalidParameter if l1_ratios is empty, any ratio is outside [0, 1], an auto-grid l1_ratio is 0, cv < 2, n_alphas == 0, or X/Y contains NaN/infinity.
• FerroError::InsufficientSamples if n_samples < cv.

type Fitted = FittedMultiTaskElasticNetCV<F>

The fitted model type returned by fit.

type Error = FerroError

The error type returned by fit.