GNNEstimator

Used to perform training and inference using a relational graph neural network (relational GNN). Every time you execute fit, predict, fit_predict, or score using a GNNEstimator object, a job is created and queued for execution. Only one job can be executed at a time. You can use JobManager.show_jobs() to check the status of all jobs.

Parameters

Name	Type	Description	Optional
connector	SnowflakeConnector	The connector object used for sending requests to the GNN engine.	No
config	GNNConfig	Configuration object containing all training hyperparameters.	No

Returns

An instance of the GNNEstimator class.

Example

from relationalai_predictive import ExperimentConfig, GNNConfig, GNNEstimator

experiment_config = ExperimentConfig(database="database_name",
                                     schema="schema_name")

gnn_config = GNNConfig(
    connector=connector,
    experiment_config=experiment_config,
    device="cuda",
    n_epochs=10,
    patience=5,
    max_iters=2000,
    text_embedder="model2vec-potion-base-4M")

estimator = GNNEstimator(connector=connector, config=gnn_config)

Methods

• fit()
• predict()
• fit_predict()
• score()