Trait Structural 

pub trait Structural: StreamingLearner {
    // Required methods
    fn apply_structural_change(&mut self, depth_delta: i32, steps_delta: i32);
    fn replacement_count(&self) -> u64;

    // Provided methods
    fn check_proactive_prune(&mut self) -> bool { ... }
    fn set_prune_half_life(&mut self, _hl: usize) { ... }
    fn tree_structure(&self) -> Vec<(usize, usize, f64, f64, u64)> { ... }
}

Available on crate feature alloc only.

Models whose internal capacity can grow or shrink at runtime.

Implemented by tree ensemble models: SGBT, AdaptiveRandomForest, DistributionalSGBT, BaggedSGBT, stacked ensembles that delegate to trees.

Object Safety

This trait is object-safe. Box<dyn Structural> is a legal type.

Required Methods

fn apply_structural_change(&mut self, depth_delta: i32, steps_delta: i32)

Apply depth/step changes that take effect at the next tree replacement.

• depth_delta – signed adjustment to maximum tree depth (+1, -1, 0).
• steps_delta – signed adjustment to ensemble step count (+2, -2, 0).

fn replacement_count(&self) -> u64

Total number of internal model replacements since creation or last reset.

External callers use this counter to detect replacement boundaries and apply queued structural changes.

Provided Methods

fn check_proactive_prune(&mut self) -> bool

Manually trigger a proactive prune check.

Returns true if an internal component was pruned or replaced. Defaults to false (no-op) for models without proactive pruning.

fn set_prune_half_life(&mut self, _hl: usize)

Dynamically set the contribution-accuracy EWMA half-life.

Recomputes prune_alpha so each correction batch contributes equally regardless of batch size. Default: no-op for models without an EWMA prune accumulator.

fn tree_structure(&self) -> Vec<(usize, usize, f64, f64, u64)>

Per-tree structure diagnostics.

Returns one tuple per tree: (max_depth, n_leaves, leaf_weight_mean, leaf_weight_std, samples_seen). Defaults to an empty vec for models without tree diagnostics.

Implementors