Pyod

PyOD is an open source Python library for anomaly and outlier detection, positioned by its maintainers as the hub library and API standard for the field. It offers more than 60 detection algorithms behind one interface and covers five data types: tabular, time series, graph, text and image data. The project also includes an orchestration core and a natural language investigation layer for working through detection results. It is written for data scientists and machine learning engineers who need to flag unusual records in fraud, monitoring, security or quality control data without learning a different API for each algorithm. The site presents PyOD as an ecosystem of four parts: the library itself, a knowledge base of curated books, papers, courses, benchmark datasets and toolboxes, a contributor community, and the thousands of public projects that depend on it. Sibling libraries such as PyGOD and TODS extend the same interface to other data types. The maintainers report that the library has been maintained since 2018 and has had more than 100 released versions. The home page cites its inclusion in tooling lists published by defense agencies and in studies by national laboratories. PyOD is free to install and use, and the site publishes no paid plans or pricing.