nexus-stats-detection

Advanced change detection and signal analysis for nexus-stats.

Detection Types

• MosumF64 — Moving sum change detection (requires alloc)
• ShiryaevRobertsF64 — Shiryaev-Roberts change detection (requires std or libm)
• AdaptiveThresholdF64 — EMA + Welford adaptive threshold (requires std or libm)
• RobustZScoreF64 — Median-based robust Z-score anomaly detection
• TrendAlertF64 — Holt-based trend alerting
• MultiGateF64 — Multi-threshold gating
• PageHinkleyF64 — Sequential mean drift detection (Page-Hinkley test)
• AdwinF64 — Adaptive window distribution change detection (requires alloc + (std or libm))

Signal Types

• AutocorrelationF64 — Lagged autocorrelation (requires alloc)
• CrossCorrelationF64 — Lagged cross-correlation (requires alloc)
• EntropyF64 — Shannon entropy estimation (requires alloc + (std or libm))
• TransferEntropyF64 — Transfer entropy between series (requires alloc + (std or libm))
• PredictiveInfoBoundF64 — Binned mutual information estimation (requires alloc + (std or libm))

Estimation Types

• SprtBernoulli — Sequential Probability Ratio Test (requires std or libm)
• SprtGaussian — Gaussian SPRT (requires std or libm)

Usage

Enable the detection feature on nexus-stats for unified import paths:

use nexus_stats::detection::RobustZScoreF64;
use nexus_stats::signal::AutocorrelationF64;

License

Licensed under either of Apache License, Version 2.0 or MIT license at your option.