Struct AnomalyDetector 

pub struct AnomalyDetector { /* private fields */ }

Detects anomalies in a stream of observations using a statistical method.

This detector maintains a sliding window of recent observations, calculates the running mean and variance of features extracted from these observations, and uses a z-score to determine if a new observation is anomalous.

Implementations

impl AnomalyDetector

pub fn new(threshold: f64, window_size: usize) -> Self

Creates a new AnomalyDetector.

Arguments

• threshold - The z-score threshold. An observation with a score above this value is considered an anomaly. A common value is 2.0 or 3.0.
• window_size - The number of recent observations to include in the sliding window for calculating statistics.

pub fn update(&mut self, obs: &Observation)

Updates the detector’s statistics with a new observation.

pub fn is_anomaly(&self, obs: &Observation) -> bool

Returns true if the observation is considered anomalous.

pub fn anomaly_score(&self, obs: &Observation) -> f64

Calculates an anomaly score for the observation.

The score is the average z-score across all extracted features, representing how many standard deviations the observation is from the mean. A higher score indicates a greater anomaly.