Struct StackingClassifier 

#[non_exhaustive]
pub struct StackingClassifier { /* private fields */ }

Stacking (stacked generalization) classifier.

Trains base estimators via k-fold cross-validation, collects out-of-fold predictions as meta-features, then trains a final meta-learner on those meta-features. At predict time, base estimator predictions are fed to the meta-learner.

Examples

use scry_learn::ensemble::StackingClassifier;
use scry_learn::tree::DecisionTreeClassifier;
use scry_learn::linear::LogisticRegression;

let sc = StackingClassifier::new(
    vec![
        Box::new(DecisionTreeClassifier::new().max_depth(3)),
        Box::new(DecisionTreeClassifier::new().max_depth(7)),
    ],
    Box::new(LogisticRegression::new()),
).cv(5);

Implementations

impl StackingClassifier

pub fn new( estimators: Vec<Box<dyn EnsembleClassifier>>, final_estimator: Box<dyn EnsembleClassifier>, ) -> Self

Create a new stacking classifier.

estimators are the base learners; final_estimator is the meta-learner.

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

Set the number of CV folds (default: 5).

pub fn seed(self, s: u64) -> Self

Set the random seed for fold generation (default: 42).

pub fn fit(&mut self, data: &Dataset) -> Result<()>

Fit the stacking classifier.

1. Split data into cv folds.
2. For each fold, train base learners on training folds and collect out-of-fold predictions.
3. Assemble meta-features from out-of-fold predictions.
4. Train the final estimator on meta-features.
5. Re-train all base learners on the full dataset for prediction.

pub fn predict(&self, features: &[Vec<f64>]) -> Result<Vec<f64>>

Predict class labels using the stacking ensemble.

Gets predictions from all base learners, then feeds them to the meta-learner.

Trait Implementations

impl Clone for StackingClassifier

fn clone(&self) -> StackingClassifier

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for StackingClassifier

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

Formats the value using the given formatter.