graphdatascience

library1.22pypypi✓ verified 83d ago

A Python client for the Neo4j Graph Data Science (GDS) library, enabling users to run graph algorithms and manage graph projections directly from Python. Current version is 1.22, requiring Python >=3.10. The library is actively maintained with regular releases aligned with GDS server versions.

database ai-ml

pip install graphdatascience

github ↗homepage ↗

INSTALL

IMPORT

SIG · GRAPHDATASCIENCE

graphdatascience

databasepythonv1.22

harness data pending

Install & Compatibility

Where this runs

No compatibility data collected yet for this library.

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

GraphDataScience

✓ from graphdatascience import GraphDataScience

✗ from graphdatascience.client import GraphDataScience

In v1.x, the main class is exposed at the package level.

Connects to a Neo4j database and retrieves the GDS library version.

import os from graphdatascience import GraphDataScience URI = os.environ.get('NEO4J_URI', 'bolt://localhost:7687') AUTH = (os.environ.get('NEO4J_USER', 'neo4j'), os.environ.get('NEO4J_PASSWORD', 'password')) gds = GraphDataScience(URI, auth=AUTH) # Run a simple algorithm result = gds.run_cypher("RETURN gds.version() AS version") print(result) gds.close()

Debug

Known issues

breakingThe 'graphdatascience' package dropped support for Python 3.7 and 3.8 in v1.0. Python 3.9 was deprecated in v1.14 and removed in v1.17. Ensure your environment uses Python >=3.10.

fix

Upgrade Python to 3.10 or later before installing current versions.

affects: >=1.0

deprecatedThe `GraphDataScience` constructor no longer supports the `auth` parameter as a tuple; pass it as a `neo4j.Auth` object or use the `gds.auth` property. The tuple form still works but is deprecated.

fix

Use `Auth(('neo4j', password))` or set `gds.auth = ('neo4j', password)` after creation.

affects: >=1.18

gotchaAlways call `gds.close()` to release connections. Forgetting this can cause resource leaks and connection exhaustion.

fix

Use context manager: `with GraphDataScience(...) as gds: ...` or ensure `.close()` is called in a `finally` block.

affects: all

gotchaWhen using `gds.run_cypher()` with queries that return large results, the client may hold all data in memory. For large graph projections, use streaming or limit results.

fix

Use `yield` or cursor-based iteration if available, or apply LIMIT in your Cypher query.

affects: all

Errors

Common errors & fixes

AttributeError: module 'graphdatascience' has no attribute 'GraphDataScience'

Incorrect import path; GraphDataScience is not a direct attribute in older versions or submodules.

fix

Use `from graphdatascience import GraphDataScience` (or `GraphDataScience` is a top-level class).

neo4j.exceptions.ServiceUnavailable: Failed to establish connection to ...

Neo4j server is not running, URI is wrong, or port is blocked.

fix

Ensure Neo4j is started and accessible at the URI. Check firewall and port 7687 (default).

ValueError: authentication credentials must be a tuple or an Auth object

Passing `auth` as a single string or incorrect format.

fix

Use `auth=('neo4j', 'password')` or `auth=Auth(('neo4j', 'password'))`.

Upgrade

Version history

1.22latest on PyPI · released Jun 4, 2026

Audit

Dependencies

neo4joptionalRequired for database connectivity

Agent activity

12 hits · last 30 days

node

OpenAI (training)

Resources

githubhttps://github.com/neo4j/graph-data-science-client ↗

homepagehttps://github.com/neo4j/graph-data-science-client ↗