Struct GradientBoostingRegressor 

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

Gradient Boosting for regression.

Builds an additive ensemble of shallow decision trees, each fitting the negative gradient (pseudo-residuals) of the loss function. Supports stochastic subsampling and multiple loss functions.

Example

use scry_learn::dataset::Dataset;
use scry_learn::tree::GradientBoostingRegressor;

let features = vec![vec![1.0, 2.0, 3.0, 4.0, 5.0]];
let target = vec![2.0, 4.0, 6.0, 8.0, 10.0];
let data = Dataset::new(features, target, vec!["x".into()], "y");

let mut gbr = GradientBoostingRegressor::new()
    .n_estimators(50)
    .learning_rate(0.1)
    .max_depth(3);
gbr.fit(&data).unwrap();

let preds = gbr.predict(&[vec![3.0]]).unwrap();
assert!((preds[0] - 6.0).abs() < 1.0);

Implementations

impl GradientBoostingRegressor

pub fn new() -> Self

Create a new regressor with default parameters.

pub fn n_estimators(self, n: usize) -> Self

Set number of boosting rounds.

pub fn learning_rate(self, lr: f64) -> Self

Set learning rate (shrinkage). Lower values need more estimators.

pub fn max_depth(self, d: usize) -> Self

Set maximum depth per tree (default: 3, shallow stumps).

pub fn min_samples_split(self, n: usize) -> Self

Set minimum samples required to split an internal node.

pub fn min_samples_leaf(self, n: usize) -> Self

Set minimum samples required in a leaf node.

pub fn subsample(self, s: f64) -> Self

Set subsample fraction (0.0, 1.0] for stochastic GBT.

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

Set random seed.

pub fn n_iter_no_change(self, n: usize) -> Self

Enable early stopping. Training stops when validation loss does not improve for n consecutive rounds.

pub fn validation_fraction(self, frac: f64) -> Self

Set fraction of training data to use as validation for early stopping (default: 0.1).

pub fn tol(self, t: f64) -> Self

Set tolerance for early stopping (default: 1e-4).

pub fn callback(self, cb: Box<dyn TrainingCallback>) -> Self

Add a training callback (invoked after each boosting round).

pub fn n_estimators_used(&self) -> usize

Number of estimators actually used (may be less than n_estimators if early stopping triggered).

pub fn loss(self, l: RegressionLoss) -> Self

Set the regression loss function.

Example

use scry_learn::tree::{GradientBoostingRegressor, RegressionLoss};

let gbr = GradientBoostingRegressor::new()
    .loss(RegressionLoss::Huber { alpha: 0.9 });

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

Train the gradient boosting ensemble.

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

Predict values for new samples.

features is row-major: features[sample_idx][feature_idx].

pub fn feature_importances(&self) -> Result<Vec<f64>>

Feature importances averaged across all trees.

pub fn n_trees(&self) -> usize

Number of estimators (trees) in the ensemble.

pub fn early_stopped(&self) -> bool

Whether early stopping was triggered.

pub fn history(&self) -> Option<&TrainingHistory>

Return training history (populated after fit()).

pub fn trees(&self) -> &[DecisionTreeRegressor]

Get individual trees (for inspection or ONNX export).

pub fn n_features(&self) -> usize

Number of features the model was trained on.

pub fn learning_rate_val(&self) -> f64

Learning rate value.

pub fn init_prediction_val(&self) -> f64

Initial (base) prediction value.

Trait Implementations

impl Clone for GradientBoostingRegressor

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Default for GradientBoostingRegressor

fn default() -> Self

Returns the “default value” for a type.

impl PipelineModel for GradientBoostingRegressor

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

Train the model on a dataset.

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

Predict on row-major feature matrix.

impl Tunable for GradientBoostingRegressor

fn set_param(&mut self, name: &str, _value: ParamValue) -> Result<()>

Apply a named hyperparameter.

fn clone_box(&self) -> Box<dyn Tunable>

Clone this model into a boxed trait object.

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

Train on a dataset.

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

Predict on row-major features.