Struct BocpdDetector

pub struct BocpdDetector {
    config: BocpdConfig,
    entity_id: u64,
    ew_mean: Vec<f64>,
    ew_var: Vec<f64>,
    prev_mean: Vec<f32>,
    count: usize,
    since_last_detection: usize,
    dim: Option<usize>,
}

Online streaming change point detector.

Fields

config: BocpdConfig

entity_id: u64

ew_mean: Vec<f64>

Exponentially weighted mean per dimension.

ew_var: Vec<f64>

Exponentially weighted variance per dimension.

prev_mean: Vec<f32>

Previous mean (for drift vector computation).

count: usize

Number of observations processed.

since_last_detection: usize

Observations since last detection (cooldown counter).

dim: Option<usize>

Dimensionality.

Implementations

impl BocpdDetector

pub fn new(entity_id: u64, config: BocpdConfig) -> Self

Create a new detector for the given entity.

pub fn observe(&mut self, timestamp: i64, vector: &[f32]) -> Option<ChangePoint>

Process a single observation.

Returns Some(ChangePoint) if a change is detected.

fn mahalanobis_like(&self, vector: &[f32]) -> f64

Compute a Mahalanobis-like distance (average z-score across dimensions).

pub fn count(&self) -> usize

Current number of observations processed.