Struct DriftDetector 

pub struct DriftDetector { /* private fields */ }

Drift detector using statistical distance measures.

Examples

use aprender::metrics::drift::{DriftDetector, DriftConfig, DriftStatus};
use aprender::primitives::Vector;

let reference = Vector::from_slice(&[1.0, 2.0, 3.0, 4.0, 5.0]);
let detector = DriftDetector::new(DriftConfig::default());

// Similar distribution - no drift
let current = Vector::from_slice(&[1.1, 2.1, 3.0, 4.0, 5.1]);
let status = detector.detect_univariate(&reference, &current);
assert!(matches!(status, DriftStatus::NoDrift | DriftStatus::Warning { .. }));

Implementations

impl DriftDetector

pub fn new(config: DriftConfig) -> Self

Create a new drift detector.

pub fn detect_univariate( &self, reference: &Vector<f32>, current: &Vector<f32>, ) -> DriftStatus

Detect drift in univariate data using normalized mean difference.

Uses (|mean_ref - mean_cur| / std_ref) as drift measure.

pub fn detect_multivariate( &self, reference: &Matrix<f32>, current: &Matrix<f32>, ) -> (DriftStatus, Vec<DriftStatus>)

Detect drift in multivariate data using feature-wise analysis.

Returns drift status for each feature and overall status.

pub fn detect_performance_drift( &self, baseline_scores: &[f32], current_scores: &[f32], ) -> DriftStatus

Detect performance drift using accuracy/score degradation.

Compares baseline performance to current performance.

Trait Implementations

impl Debug for DriftDetector

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

Formats the value using the given formatter.