Module changepoint 

Changepoint Detection for Pattern Monitoring

Implements changepoint detection using the PELT (Pruned Exact Linear Time) algorithm from the augurs-changepoint crate. This module detects significant changes in pattern metrics over time, enabling adaptive pattern learning.

Example

use do_memory_core::patterns::changepoint::{ChangepointDetector, ChangepointConfig};

// Create detector with default settings (needs to be mutable for detection)
let mut detector = ChangepointDetector::new(ChangepointConfig::default());

// Simulate pattern success rate time series
let metrics = vec![
    0.8, 0.82, 0.81, 0.79, 0.83, // Normal variation
    0.45, 0.48, 0.42, 0.44,      // Drop (changepoint detected)
    0.46, 0.47, 0.45, 0.48,      // New baseline
];

// Detect changepoints
let changepoints = detector.detect_changepoints(&metrics).unwrap();
println!("Detected {} changepoints", changepoints.len());

Integration with Monitoring

The changepoint detector integrates with the agent monitoring system to:

• Detect significant changes in pattern success rates
• Identify shifts in task execution metrics
• Trigger pattern recalibration when drift is detected

Re-exports

pub use detector::ChangepointDetector;

pub use types::ChangeDirection;

pub use types::ChangeType;

pub use types::Changepoint;

pub use types::ChangepointConfig;

pub use types::ChangepointError;

pub use types::SegmentComparison;

pub use types::SegmentComparisonConfig;

pub use types::SegmentStats;

Modules

algorithms

Changepoint Algorithms

detector

Changepoint Detector

tests

Changepoint Tests

types

Changepoint Types