cusum

Twitter's Anomaly Detection in Pure Python

Easier CUSUM control charts. Returns simple CUSUM statistics, CUSUMs with control limit calculations, and function to generate faceted CUSUM Control Charts

Different flavours of CUSUM for change point detection.

Calibrated simulation and detectors for MES-embedded carbon-intensity monitoring of energy anomalies in machining-style processes. Characterizes the adaptive-baseline inertia blind spot and proposes an event-anchored + residual-CUSUM detector that closes it. Reproducible: code, raw sweep data, and figure scripts.

Statistical volume anomaly detection for trade streams - Hawkes process, CUSUM, and Bayesian Online Changepoint Detection (BOCPD). Zero dependencies. TypeScript.

IMU fault detection

Fast Online Changepoint Detection via Functional Pruning CUSUM statistics

NASA Bearing Dataset: Fault Detection with Wiener denoising and custom time-frequency btstft Transforms

Social Networks Monitoring

Quickest Change Detection for Unnormalized Statistical Models

Calibrated online CUSUM and EWMA change detection with held-out false-alarm audits

Anomaly Detection in Sensor Data (LIT101) from Secure Water Treatment (SWaT) testbed . Demo of CUSUM and MLP methodologies.

This repository represents additional control charts, various plans and variables that are used within the chart scope using Minitab software

CUSUM is the cumulative sum of the samples and CUMEAN is the cumulative sum of the updated samples with their mean. CUSUM and CUMEAN can detect relatively small changes in a process mean. They can be more useful in the time series dataset.

Regime detection without religion — six algorithms (HMM, BOCPD, CUSUM, GMM, BinSeg, Ensemble), one harness, reproducible leaderboard.

Changepoint detection toolkit for offline and online in Rust with Python bindings

NCIs Project 2024/25

Adversarial game-integrity system that detects rigged casino games and advantage players using sequential hypothesis testing (SPRT/CUSUM) and anomaly detection.

A robust Federated Learning framework implementing the novel FedCADS-UCB algorithm. Engineered for resilient client selection using CUSUM drift detection, adaptive multi-armed bandits, and hierarchical clustering to maintain high accuracy (>95%) during concept drift and label poisoning attacks.

Streaming anomaly detection in Rust - detectors, calibration, SOC triage. Powers eBPFsentinel