Struct FittedMultiLabelBinarizer 

pub struct FittedMultiLabelBinarizer { /* private fields */ }

A fitted multi-label binarizer holding the discovered class set.

Created by calling Fit::fit on a MultiLabelBinarizer.

Implementations

impl FittedMultiLabelBinarizer

pub fn classes(&self) -> &[usize]

Return the sorted class labels discovered during fitting.

pub fn n_classes(&self) -> usize

Return the number of unique classes.

pub fn inverse_transform( &self, y: &Array2<f64>, ) -> Result<Vec<Vec<usize>>, FerroError>

Map a multi-hot indicator matrix back to label sets.

The indicator matrix must contain only exact 0.0 and 1.0 values; a class is included for a sample iff its cell is exactly 1.0. This mirrors scikit-learn 1.5.2 MultiLabelBinarizer.inverse_transform (sklearn/preprocessing/_label.py:941-947), which validates the matrix with np.setdiff1d(yt, [0, 1]) and raises ValueError on any value outside {0, 1} before selecting classes where the cell == 1.

Errors

Returns FerroError::ShapeMismatch if the number of columns does not match the number of classes. Returns FerroError::InvalidParameter if any cell value is not exactly 0.0 or 1.0.

Trait Implementations

impl Clone for FittedMultiLabelBinarizer

fn clone(&self) -> FittedMultiLabelBinarizer

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for FittedMultiLabelBinarizer

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

Formats the value using the given formatter.

impl Transform<Vec<Vec<usize>>> for FittedMultiLabelBinarizer

fn transform(&self, y: &Vec<Vec<usize>>) -> Result<Array2<f64>, FerroError>

Transform label sets into a multi-hot indicator matrix.

Each row has a 1.0 in every column corresponding to one of its labels and 0.0 elsewhere.

Labels not seen during fitting are silently ignored: the indicator is built only from known labels (mirroring scikit-learn 1.5.2 MultiLabelBinarizer._transform, sklearn/preprocessing/_label.py:889-902). scikit-learn additionally emits a warnings.warn("unknown class(es) ... will be ignored"); that warning is intentionally not emitted here because the crate has no logging facade and adding one would be out of scope.

The Result return type is retained because the Transform trait requires it; transform always returns Ok.

type Output = ArrayBase<OwnedRepr<f64>, Dim<[usize; 2]>>

The transformed output type.

type Error = FerroError

The error type returned by transform.