scientific-anomaly-detection

What problem does it solve?

This skill provides methods to detect anomalies, outliers, and unusual patterns in scientific data, using ensemble techniques (Isolation Forest, LOF, One-Class SVM) and Autoencoders, complemented by SPC-based process monitoring and multivariate anomaly detection.

Core Features & Use Cases

• Ensemble anomaly detection: combining multiple methods to robustly identify outliers.
• Autoencoder-based anomaly detection for complex patterns.
• SPC and multivariate anomaly scoring with threshold optimization.
• OpenML integration for benchmarking and datasets.

Quick Start

Run anomaly detection on your dataset using the ensemble workflow to obtain anomaly scores and flagged outliers.