Enum ClassWeight 

pub enum ClassWeight<F> {
    None,
    Balanced,
    Explicit(Vec<(usize, F)>),
}

Per-class weighting strategy for LinearSVC.

Mirrors sklearn.svm.LinearSVC’s class_weight parameter (sklearn/svm/_classes.py:118-124, constraint {None, dict, 'balanced'}): it scales the inverse-regularization C per class so the effective penalty for class i is class_weight[i]·C (compute_class_weight, sklearn/svm/_base.py:1179; weighted_C[i] = C·class_weight[i], liblinear/linear.cpp:2496-2507). The expanded per-class weights are computed by [compute_class_weight] following sklearn.utils.compute_class_weight semantics (sklearn/utils/class_weight.py:63-81).

This mirrors ferrolearn_linear::sgd::ClassWeight for cross-estimator consistency, but is defined locally (no cross-import of sgd internals).

Variants

None

Uniform weights (all classes weighted 1.0). The default (class_weight=None, class_weight.py:63-65).

Balanced

Balanced weights n_samples / (n_classes · count_c) per class c, matching sklearn.utils.compute_class_weight("balanced", ...) (class_weight.py:66-74).

Explicit(Vec<(usize, F)>)

Explicit class-label -> weight map. Classes absent from the map default to 1.0, matching the dict branch of compute_class_weight (class_weight.py:75-81).

Trait Implementations

impl<F: Clone> Clone for ClassWeight<F>

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

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<F: Debug> Debug for ClassWeight<F>

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

Formats the value using the given formatter.

impl<F> Default for ClassWeight<F>

fn default() -> ClassWeight<F>

Returns the “default value” for a type.