TabularDataset

Wraps a connector and a TabularTask for use with ICLEstimator. Unlike RelationalDataset, there are no table declarations, foreign keys, or candidate keys — the task’s table paths (context_table, test_table, validation_table) are the only data source, and no DFS/cross-table feature generation happens before inference.

Parameters

Parameter	Type	Description	Optional
connector	SnowflakeConnector	Connection to the data backend (used for auth / path resolution).	No
name	str	Dataset name (used for logging and job naming).	No
task	TabularTask	Describes the splits, label column, and task type.	No

Returns

An instance of the TabularDataset class.

Example

from relationalai_predictive import TabularTask, TabularDataset, TaskType

task = TabularTask(
    name="churn",
    task_type=TaskType.BINARY_CLASSIFICATION,
    label_column="churned",
    context_table="DATABASE.SCHEMA.CHURN_CONTEXT",
    test_table="DATABASE.SCHEMA.CHURN_TEST",
)

dataset = TabularDataset(connector=connector, name="churn", task=task)

Pass the resulting dataset directly to ICLEstimator.predict() or ICLEstimator.score().