graphdatascience

What it is and what it does

graphdatascience is a Python wrapper around Neo4j's Graph Data Science library, designed to let you work with graph algorithms and machine learning pipelines without writing Cypher. It abstracts the Neo4j Python driver and provides object-oriented access to graph projections, algorithm execution, and model training. The client mimics GDS's Cypher procedure API but surfaces it as Python methods, reducing boilerplate and making graph workflows feel more native to Python developers.

You use it to load or project graphs from a Neo4j database, run algorithms like PageRank or community detection, and construct machine learning pipelines for tasks like node classification or link prediction. It depends on neo4j (the driver), pandas, numpy, and several utility libraries (pydantic, tenacity, requests) to handle data transformation, retry logic, and communication. The package is production-stable and actively maintained, with support for Python 3.10 through 3.14.

Use it for:

• Build and train node classification pipelines on graph data without writing Cypher queries.
• Run graph algorithms (PageRank, centrality, community detection) and materialize results back to the database.
• Develop recommendation systems using graph embeddings and similarity algorithms like kNN on FastRP.
• Perform end-to-end machine learning on heterogeneous graphs with feature engineering and model inference.
• Load and manipulate graph projections programmatically as part of a Python data science workflow.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

A Python client for Neo4j Graph Data Science that lets you project graphs, run algorithms, and build machine learning pipelines using pure Python code instead of Cypher.

Yes, if you are already using Neo4j Graph Data Science and want to work with graphs from Python. The low install friction, active maintenance, permissive license, and lack of known vulnerabilities make it a safe choice. No if you don't have a GDS instance or are looking for a standalone graph library—this is a client, not a self-contained engine.

Install

graphdatascience on PyPI

pip install graphdatascience

uv add graphdatascience

poetry add graphdatascience

Installing graphdatascience

Before you install

Low friction: pure Python wheel with no compiled dependencies. Active maintenance with recent releases; last commit 2026-08-13. Requires Python 3.10 or later and a running Neo4j GDS instance.

License in practice

Apache-2.0 permissive license allows commercial and private use with minimal restrictions; suitable for most projects.

Quickstart

pip install graphdatascience

from graphdatascience import GraphDataScience

gds = GraphDataScience("neo4j+s://your-instance:7687", auth=("user", "pass"), aura_ds=True)
G = gds.graph.load_cora()
result = gds.pageRank.mutate(G, tolerance=0.5, mutateProperty="pagerank")

Requires a running Neo4j Graph Data Science instance (GDS 2.0+) and network access to it; cannot operate standalone.

Verify before relying

• Whether numeric utility functions are truly unsupported or if this limitation has changed since the description was written.
• Exact compatibility matrix between graphdatascience versions and GDS server versions beyond the stated 2.0+ requirement.

Package facts

Evidence: graphdatascience-1.22-py3-none-any.whl

Tags

More Software Development packages