Struct WeightedEnsembleModel 

pub struct WeightedEnsembleModel { /* private fields */ }

Weighted Ensemble Model using weak supervision

Phase 6: Combines multiple noisy signals (SBFL, TDG, Churn, Complexity, RAG) to learn optimal weights for defect prediction.

Implementations

impl WeightedEnsembleModel

pub fn new() -> Self

Create a new ensemble model with default labeling functions

pub fn with_labeling_functions(lfs: Vec<Box<dyn LabelingFunction>>) -> Self

Create with custom labeling functions

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

Set number of EM iterations

pub fn fit(&mut self, files: &[FileFeatures]) -> Result<()>

Fit the model using EM algorithm on unlabeled data

This learns optimal weights for each labeling function by maximizing the likelihood of the observed label matrix.

pub fn predict(&self, features: &FileFeatures) -> f32

Predict defect probability for a file

pub fn get_weights(&self) -> Option<&LabelModelWeights>

Get learned weights for interpretability

pub fn is_fitted(&self) -> bool

Check if model is fitted

pub fn save(&self, path: &Path) -> Result<()>

Save model weights to file

pub fn load(&mut self, path: &Path) -> Result<()>

Load model weights from file

Trait Implementations

impl Default for WeightedEnsembleModel

fn default() -> Self

Returns the “default value” for a type.