latent.flows.drift_flow

drift_flow — Detect distribution drift between datasets (FR-3.3).

Functions

drift_flow

drift_flow(reference_scores: dict[str, np.ndarray], current_scores: dict[str, np.ndarray], score_types: dict[str, str] | None = None, threshold: float = 0.05, confidence_level: float = 0.95, seed: int | None = None) -> dict[str, Any]

Detect distribution drift between reference and current datasets.

Args: reference_scores: Dict mapping metric name to reference score arrays. current_scores: Dict mapping metric name to current score arrays. score_types: Optional dict mapping metric name to score type. threshold: p-value threshold for drift detection. confidence_level: Confidence level for CIs. seed: Random seed.

Returns: Dict with keys: drift_results, drift_detected, report_text.