GraphDataScience

class graphdatascience.GraphDataScience

Primary API class for the Neo4j Graph Data Science Python Client. Always bind this object to a variable called gds.

__init__(endpoint: str | Driver | QueryRunner, auth: tuple[str, str] | None = None, aura_ds: bool | None = None, database: str | None = None, arrow: str | bool = True, bookmarks: Bookmarks | None = None, show_progress: bool = True, arrow_client_options: dict[str, Any] | None = None)

Construct a new GraphDataScience object.

Parameters:

  • endpoint (str | Driver | QueryRunner) – The Neo4j endpoint to connect to. Most commonly, this is a Bolt connection URI.

  • auth (tuple[str, str] | None) – A username, password pair for database authentication.

  • aura_ds (bool | None) – A flag that indicates that the client is used to connect to a Neo4j AuraDS instance. If not set, the client will automatically derive whether the database is hosted in Aura.

  • database (str | None) – The Neo4j database to query against.

  • arrow (str | bool) –

    Arrow connection information. This is either a string or a bool.

    • If it is a string, it will be interpreted as a connection URL to a GDS Arrow Server.

    • If it is a bool:

      • True will make the client discover the connection URI to the GDS Arrow server via the Neo4j endpoint.

      • False will make the client use Bolt for all operations.

  • bookmarks (Bookmarks | None) – The Neo4j bookmarks to require a certain state before the next query gets executed.

  • show_progress (bool) – A flag to indicate whether to show progress bars for running procedures.

  • arrow_client_options (dict[str, Any] | None) – Additional options to be passed to the Arrow Flight client.

property all_shortest_paths: AllShortestPathEndpoints

Return endpoints for the all shortest paths algorithm.

property article_rank: ArticleRankEndpoints

Return endpoints for the article rank algorithm.

property articulation_points: ArticulationPointsEndpoints

Return endpoints for the articulation points algorithm.

property bellman_ford: SingleSourceBellmanFordEndpoints

Return endpoints for the single source Bellman-Ford algorithm.

property betweenness: BetweennessEndpoints

Return endpoints for the betweenness centrality algorithm.

property betweenness_centrality: BetweennessEndpoints

Return endpoints for the betweenness centrality algorithm.

property bfs: BFSEndpoints

Return endpoints for the Breadth First Search (BFS) algorithm.

bookmarks() → Bookmarks | None

Get the Neo4j bookmarks defining the currently required states for queries to execute

Returns:

The Neo4j bookmarks defining the currently required state.

Return type:

neo4j.Bookmarks | None

property bridges: BridgesEndpoints

Return endpoints for the bridges algorithm.

property clique_counting: CliqueCountingEndpoints

Return endpoints for the clique counting algorithm.

close() → None

Close the GraphDataScience object and release any resources held by it.

If the GraphDataScience object was instantiated with a Neo4j Driver, the driver will not be closed as we cannot assume sole ownership of it.

Return type:

None

property closeness: ClosenessEndpoints

Return endpoints for the closeness centrality algorithm.

property closeness_centrality: ClosenessEndpoints

Return endpoints for the closeness centrality algorithm.

property collapse_path: CollapsePathEndpoints

Return endpoints for collapsing relationship paths.

property conductance: ConductanceEndpoints

Return endpoints for the conductance algorithm.

property config: ConfigEndpoints

Return endpoints for configuration.

property dag: DagCypherEndpoints

Return endpoints for Directed Acyclic Graph (DAG) algorithms.

database() → str | None

Get the database which queries are run against.

Returns:

The name of the database.

Return type:

str | None

db_driver() → Driver

Get the Neo4j driver used by this client to communicate with the Neo4j DBMS.

This is mainly useful when the run_cypher() API is too simple for a use case, as the driver allows full control over sessions and transactions.

The driver is closed by close() if the client created it itself, after which it is unusable; a driver supplied at construction is never closed by this client.

Returns:

The Neo4j driver used by this client.

Return type:

neo4j.Driver

property debug: DebugEndpoints

Return debug endpoints.

property degree: DegreeEndpoints

Return endpoints for the degree centrality algorithm.

property degree_centrality: DegreeEndpoints

Return endpoints for the degree centrality algorithm.

property dfs: DFSEndpoints

Return endpoints for the Depth First Search (DFS) algorithm.

driver_config() → dict[str, Any]

Get the configuration used to create the underlying driver used to make queries to Neo4j.

Returns:

The configuration as a dictionary.

Return type:

dict[str, Any]

property eigenvector: EigenvectorEndpoints

Return endpoints for the eigenvector centrality algorithm.

property eigenvector_centrality: EigenvectorEndpoints

Return endpoints for the eigenvector centrality algorithm.

property fast_rp: FastRPEndpoints

Return endpoints for the fast RP algorithm.

property graph: CatalogCypherEndpoints

Return endpoints for graph management.

property graph_sage: GraphSageEndpoints

Return endpoints for the GraphSage algorithm.

property harmonic: ClosenessHarmonicEndpoints

Return endpoints for the harmonic centrality algorithm.

property harmonic_centrality: ClosenessHarmonicEndpoints

Return endpoints for the harmonic centrality algorithm.

property hash_gnn: HashGNNEndpoints

Return endpoints for the HashGNN algorithm.

property hdbscan: HdbscanEndpoints

Return endpoints for the HDBSCAN algorithm.

property hits: HitsEndpoints

Return endpoints for the HITS algorithm.

property influence_maximization_celf: CelfEndpoints

Return endpoints for the influence maximization CELF algorithm.

is_licensed() → bool

Return whether the installed GDS library is licensed.

Return type:

bool

property k1_coloring: K1ColoringEndpoints

Return endpoints for the K1 coloring algorithm.

property k_core_decomposition: KCoreEndpoints

Return endpoints for the K-core decomposition algorithm.

property k_spanning_tree: KSpanningTreeEndpoints

Return endpoints for the K-spanning tree algorithm.

property kcore: KCoreEndpoints

Return endpoints for the K-core decomposition algorithm.

property kge: KgeEndpoints

Return endpoints for KGE (TransE/DistMult) relationship prediction.

property kmeans: KMeansEndpoints

Return endpoints for the K-means algorithm.

property knn: KnnEndpoints

Return endpoints for the K-nearest neighbors algorithm.

property label_propagation: LabelPropagationEndpoints

Return endpoints for the label propagation algorithm.

last_bookmarks() → Bookmarks | None

Get the Neo4j bookmarks defining the state following the most recently called query

Returns:

The Neo4j bookmarks defining the state following the most recently called query.

Return type:

neo4j.Bookmarks | None

property leiden: LeidenEndpoints

Return endpoints for the Leiden algorithm.

property license: LicenseEndpoints

Return license endpoints.

property list_progress: ListProgressCypherEndpoint

Return endpoint for listing progress.

property local_clustering_coefficient: LocalClusteringCoefficientEndpoints

Return endpoints for the local clustering coefficient algorithm.

property louvain: LouvainEndpoints

Return endpoints for the Louvain algorithm.

property max_flow: MaxFlowEndpoints

Return endpoints for the Max Flow algorithm.

property max_k_cut: MaxKCutEndpoints

Return endpoints for the Max K-cut algorithm.

property memory: MemoryEndpoints

Return memory usage endpoints.

property model: ModelCatalogEndpoints

Return model-related endpoints for model management.

property modularity: ModularityEndpoints

Return endpoints for the modularity algorithm.

property modularity_optimization: ModularityOptimizationEndpoints

Return endpoints for the modularity optimization algorithm.

property node2vec: Node2VecEndpoints

Return endpoints for the Node2Vec algorithm.

property node_similarity: NodeSimilarityEndpoints

Return endpoints for the node similarity algorithm.

property page_rank: PageRankEndpoints

Return endpoints for the PageRank algorithm.

property pipeline: PipelineEndpoints

Return endpoints for pipeline procedures.

property prize_steiner_tree: PrizeSteinerTreeEndpoints

Return endpoints for the prize-collecting Steiner tree algorithm.

property random_walk: RandomWalkEndpoints

Return endpoints for the Random Walk algorithm.

run_cypher(query: str, params: dict[str, Any] | None = None, database: str | None = None, mode: QueryMode | Literal['READ', 'WRITE'] = QueryMode.WRITE, auto_commit: bool = False) → DataFrame

Run a Cypher query

Parameters:

  • query (str) – the Cypher query

  • params (dict[str, Any]) – parameters to the query

  • database (str) – the database on which to run the query

  • mode (QueryMode | Literal['READ', 'WRITE']) – the query mode to use. Set based on the operation performed in the query.

  • auto_commit (bool) – run the query in an auto-commit transaction. This is required for queries using CALL { … } IN TRANSACTIONS.

Returns:

The query result as a DataFrame

Return type:

pandas.DataFrame

property scale_properties: ScalePropertiesEndpoints

Return endpoints for scaling node properties.

property scc: SccEndpoints

Return endpoints for the strongly connected components algorithm.

set_bookmarks(bookmarks: Bookmarks | None) → None

Set Neo4j bookmarks to require a certain state before the next query gets executed

Parameters:

bookmarks (neo4j.Bookmarks | None) – The Neo4j bookmarks defining the required state

Return type:

None

set_database(database: str) → None

Set the database which queries are run against.

Parameters:

database (str) – The name of the database to run queries against.

Return type:

None

set_show_progress(show_progress: bool) → None

Set whether to show progress for running procedures.

Parameters:

show_progress (bool) – Whether to show progress for procedures.

Return type:

None

property shortest_path: ShortestPathEndpoints

Return endpoints for the shortest path algorithm.

property similarity: SimilarityFunctions

Return similarity functions computed client-side.

property sllpa: SllpaEndpoints

Return endpoints for the speaker-listener label propagation algorithm.

property spanning_tree: SpanningTreeEndpoints

Return endpoints for the spanning tree algorithm.

property steiner_tree: SteinerTreeEndpoints

Return endpoints for the Steiner tree algorithm.

Return endpoints for topological link prediction functions.

property triangle_count: TriangleCountEndpoints

Return endpoints for the triangle count algorithm.

property triangles: TrianglesEndpoints

Return endpoint for the triangles algorithm.

property util: UtilEndpoints

Return utility endpoints.

property wcc: WccEndpoints

Return endpoints for the weakly connected components algorithm.