Leverage all the Galileo ‘building blocks’ that are logged and stored for you to create Tests using Galileo Conditions — a class for building custom data quality checks.

Conditions are simple and flexible, allowing you to author powerful data/model tests.

​
Run Report

Integrate with email or slack to automatically receive a report of Condition outcomes after a run finishes processing.

​
Examples


    Example 1: Alert if over 50% of high DEP (>=0.7) data contains PII

        >>> c = Condition(
        ...     operator=Operator.gt,
        ...     threshold=0.5,
        ...     agg=AggregateFunction.pct,
        ...     filters=[
        ...         ConditionFilter(
        ...             metric="data_error_potential", operator=Operator.gte, value=0.7
        ...         ),
        ...         ConditionFilter(
        ...             metric="galileo_pii", operator=Operator.neq, value="None"
        ...         ),
        ...     ],
        ... )
        >>> dq.register_run_report(conditions=[c])


    Example 2: Alert if at least 20% of the dataset has drifted (Inference DataFrames only)

        >>> c = Condition(
        ...     operator=Operator.gte,
        ...     threshold=0.2,
        ...     agg=AggregateFunction.pct,
        ...     filters=[
        ...         ConditionFilter(
        ...             metric="is_drifted", operator=Operator.eq, value=True
        ...         ),
        ...     ],
        ... )
        >>> dq.register_run_report(conditions=[c])

Get started building your own Reports with Galileo Conditions

Was this page helpful?

Suggest editsRaise issue
Build your own conditionsMulti Label Text Classification