pygod-team/pygod

Health v2 · maintenance only

Adoption not part of the score

1496 stars · 139 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

PyGOD is a Python library for graph outlier detection (anomaly detection) built on PyTorch and PyTorch Geometric. It provides 10+ graph-based anomaly detection algorithms with a unified API following the PyOD design, supporting node-, edge-, and graph-level detection.

Use cases

• detect anomalies in graph-structured data
• find fraudulent accounts in social or transaction networks
• detect suspicious nodes in security systems
• run graph neural network based outlier detection in Python
• benchmark graph anomaly detection algorithms
• scale outlier detection to large graphs with mini-batching

When to choose

• your data is naturally graph-structured and you need anomaly detection
• you want a scikit-learn/PyOD-style unified API over multiple graph detectors
• you already use PyTorch Geometric and want compatible data handling
• you need node-, edge-, or graph-level detection in one toolkit

When to avoid

• your data is tabular with no graph structure (use PyOD instead)
• you need a production serving system rather than a modeling library
• you work outside the PyTorch ecosystem, e.g. TensorFlow or JAX

Facets

library · maturity active

machine-learning data-science machine-learning security data-science python cross-platform anomaly-detection outlier-detection graph-neural-networks pytorch pytorch-geometric fraud-detection graph-mining algorithms

Member repositories

For agents

markdown · JSON · MCP: product_card(name="pygod-team/pygod")