Struct GradientBoostingRegressor 

pub struct GradientBoostingRegressor { /* private fields */ }

Gradient boosting for regression (squared error). Defaults match scikit-learn’s GradientBoostingRegressor: 100 stages, learning rate 0.1, depth-3 trees.

use dml_core::GradientBoostingRegressor;
let gb = GradientBoostingRegressor::new()
    .n_estimators(300)
    .learning_rate(0.05)
    .max_depth(Some(4))
    .subsample(0.8)
    .seed(1);

Implementations

impl GradientBoostingRegressor

pub fn new() -> Self

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

Number of boosting stages (default 100).

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

Shrinkage applied to every tree (default 0.1). Smaller values need more stages but usually generalise better.

pub fn max_depth(self, depth: Option<usize>) -> Self

Maximum depth of each tree (default 3); None for unlimited.

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

Minimum number of rows to split a node (default 2).

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

Minimum number of rows in each leaf (default 1).

pub fn max_features(self, m: MaxFeatures) -> Self

Features considered per split (default: all).

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

Fraction of rows drawn (without replacement) for each stage (default 1.0); below 1 gives stochastic gradient boosting.

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

Seed for row subsampling and feature subsets (default 0).

pub fn feature_importances(&self) -> Option<&Array1<f64>>

Total impurity reduction per feature over all trees (sums to one).

pub fn n_trees(&self) -> usize

Number of fitted trees.

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 Debug for GradientBoostingRegressor

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

Formats the value using the given formatter.

impl Default for GradientBoostingRegressor

fn default() -> Self

Returns the “default value” for a type.

impl Regressor for GradientBoostingRegressor

fn fit(&mut self, x: ArrayView2<'_, f64>, y: ArrayView1<'_, f64>) -> Result<()>

fn fit_weighted( &mut self, x: ArrayView2<'_, f64>, y: ArrayView1<'_, f64>, w: ArrayView1<'_, f64>, ) -> Result<()>

Fit minimising a w-weighted loss (needed by the RLearner). The default reports that weights are unsupported.

fn predict(&self, x: ArrayView2<'_, f64>) -> Result<Array1<f64>>

fn fresh(&self) -> Box<dyn Regressor>