Struct HoeffdingTreeClassifier 

pub struct HoeffdingTreeClassifier { /* private fields */ }

Available on crate feature alloc only.

A streaming decision tree classifier based on the VFDT algorithm.

Grows incrementally by maintaining per-leaf class distributions and histogram bins. Splits are committed when the Hoeffding bound guarantees the best information-gain split is statistically superior to the runner-up.

Thread Safety

HoeffdingTreeClassifier is Send + Sync, making it usable in async and multi-threaded pipelines.

Implementations

impl HoeffdingTreeClassifier

pub fn new(config: HoeffdingClassifierConfig) -> Self

Create a new classifier from the given configuration.

If config.n_features > 0, the tree is immediately initialized with a root leaf. Otherwise, initialization is deferred until the first training sample arrives.

pub fn train_one(&mut self, features: &[f64], class: usize)

Train on a single observation: route to leaf, update stats, maybe split.

Arguments

• features – feature vector for this observation.
• class – the class label (0-indexed).

pub fn predict_class(&self, features: &[f64]) -> usize

Predict the majority class for the given feature vector.

Returns the class index with the highest count at the reached leaf. If no samples have been seen, returns 0.

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

Predict the class probability distribution for the given feature vector.

Returns a Vec<f64> of length n_classes where each entry is the estimated probability (fraction of samples) for that class. The probabilities sum to 1.0 (or all zeros if no samples seen).

pub fn n_leaves(&self) -> usize

Number of leaf nodes in the tree.

pub fn n_nodes(&self) -> usize

Total number of nodes (leaves + splits) in the tree.

pub fn max_depth_seen(&self) -> usize

Maximum depth reached by any node in the tree.

pub fn n_classes(&self) -> usize

Number of discovered classes.

pub fn n_samples_seen(&self) -> u64

Total number of training samples seen since creation or last reset.

pub fn reset(&mut self)

Reset the tree to its initial (untrained) state.

Clears all nodes and counters. If n_features was configured up front, the root leaf is re-created; otherwise initialization is deferred again.

Trait Implementations

impl Clone for HoeffdingTreeClassifier

fn clone(&self) -> HoeffdingTreeClassifier

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for HoeffdingTreeClassifier

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

Formats the value using the given formatter.

impl StreamingLearner for HoeffdingTreeClassifier

fn train_one(&mut self, features: &[f64], target: f64, _weight: f64)

Train on a single observation.

The target is cast to usize for the class label. Weight is currently unused (all samples contribute equally).

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

Predict the majority class as a floating-point value.

fn n_samples_seen(&self) -> u64

Total number of observations trained on since creation or last reset.

fn reset(&mut self)

Reset the model to its initial (untrained) state.

fn train(&mut self, features: &[f64], target: f64)

Train on a single observation with unit weight.

fn train_one_weighted(&mut self, features: &[f64], target: f64, weight: f64)

Train on a single observation with an explicit distillation weight.

fn predict_batch(&self, feature_matrix: &[&[f64]]) -> Vec<f64>

Predict for each row in a feature matrix.