Struct FittedPolynomialFeatures 

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

A fitted PolynomialFeatures.

PolynomialFeatures.fit (sklearn _polynomial.py:306-400, “Compute number of output features”) learns NO numeric statistics — it records only the shape metadata derivable from n_features_in_ and the configured params: n_features_in_, n_output_features_ (the number of output columns), and the powers_ exponent matrix. The fitted type’s Transform::transform REUSES the very same combination enumeration ([PolynomialFeatures:: feature_combinations]) and value math ([generate_poly_features]) as the stateless PolynomialFeatures path, so the two paths are bit-identical; it additionally enforces the transform-time feature-count check against n_features_in_ (sklearn X has N features, but ... expecting M).

Implementations

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

pub fn n_features_in(&self) -> usize

Return the number of features (columns) seen during Fit::fit.

Mirrors scikit-learn’s PolynomialFeatures.n_features_in_ (_polynomial.py:323).

pub fn n_output_features(&self) -> usize

Return the total number of polynomial output columns.

Mirrors scikit-learn’s PolynomialFeatures.n_output_features_ (_polynomial.py:362).

pub fn powers(&self) -> &Array2<usize>

Return the powers_ exponent matrix of shape (n_output_features_, n_features_in_).

powers_[i, j] is the exponent of input feature j in output feature i, mirroring scikit-learn’s PolynomialFeatures.powers_ (_polynomial.py:250-264).

pub fn degree(&self) -> usize

Return the configured degree.

pub fn interaction_only(&self) -> bool

Return whether only interaction terms are generated.

pub fn include_bias(&self) -> bool

Return whether a bias column is included.

Trait Implementations

impl<F: Clone> Clone for FittedPolynomialFeatures<F>

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

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<F: Debug> Debug for FittedPolynomialFeatures<F>

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

Formats the value using the given formatter.

impl<F: Float + Send + Sync + 'static> Transform<ArrayBase<OwnedRepr<F>, Dim<[usize; 2]>>> for FittedPolynomialFeatures<F>

fn transform(&self, x: &Array2<F>) -> Result<Array2<F>, FerroError>

Generate polynomial features, delegating to the SAME combination enumeration + value math as the stateless PolynomialFeatures path.

First applies the REQ-8 check_array guards (min-samples / min-features / finite) and generates the polynomial matrix, THEN validates that x has the same number of columns recorded during Fit::fit. This ORDER matches sklearn’s _validate_data(reset=False), which runs check_array BEFORE the n_features_in_ consistency check (_polynomial.py:433-435, #2207): a NaN / ±inf / zero-sample / zero-feature input raises its check_array error EVEN when the column count is also wrong. Only after that does the feature-count comparison fire. Because the combinations are re-enumerated from the SAME params (degree / interaction_only / include_bias) and the value math is the shared [generate_poly_features], the output is byte-identical to PolynomialFeatures::transform.

Errors

Returns FerroError::ShapeMismatch if the column count differs from n_features_in_ (sklearn ValueError: X has N features, but PolynomialFeatures is expecting M features as input., _polynomial.py:402-435). Returns FerroError::InsufficientSamples for zero rows and FerroError::InvalidParameter for zero features or any non-finite value (REQ-8, via [validate_poly_input]).

type Output = ArrayBase<OwnedRepr<F>, Dim<[usize; 2]>>

The transformed output type.

type Error = FerroError

The error type returned by transform.