yzhao062/pyod

PyOD 3 stands as the most comprehensive Python library for anomaly detection, designed for scalability and ease of use across diverse data types. It offers a rich collection of over 60 anomaly detection algorithms, capable of processing multi-modal data including tabular, time series, graph, text, and images through a unified API. A significant advancement is its 'agentic' capability, empowering AI agents to conduct sophisticated anomaly investigations using natural language, thereby democratizing expert-level detection without requiring deep outlier detection expertise. The library supports the entire anomaly detection lifecycle, from data ingestion to anomaly explanation and guidance, all underpinned by extensive benchmarks for robust performance.