meridian.analysis.review.results.ReviewSummary

The final summary of all model quality checks.

meridian.analysis.review.results.ReviewSummary(
    overall_status: meridian.analysis.review.results.Status,
    summary_message: str,
    results: list[CheckResult],
    health_score: float,
    channel_calibration_status: Mapping[str, bool] = dataclasses.field(default_factory=dict),
    calibrated_channel_names: Sequence[str] = dataclasses.field(default_factory=list),
    implausible_roi_chart_json: (str | None) = None,
    high_variance_chart_json: (str | None) = None,
    potential_bias_chart_json: (str | None) = None,
    calibration_overview_data: Sequence[meridian.analysis.review.results.CalibrationOverviewChannelData] = ()
)

overall_status The overall status of all checks.
summary_message A summary message of all checks.
results A list of all check results.
health_score The health score of the model.
channel_calibration_status Mapping of channel name to calibration status.
calibrated_channel_names Sequence of calibrated channel names.
implausible_roi_chart_json Chart JSON for implausible ROI recommendation plot.
high_variance_chart_json Chart JSON for high variance recommendation plot.
potential_bias_chart_json Chart JSON for potential bias recommendation plot.
calibration_overview_data Sequence of calibration overview channel data.
calibration_score Computes the overall model-level calibration score (arithmetic average).
channel_calibration_recommendations Computes per-channel calibration recommendation data.

The recommendations are filtered based on the total number of channels: - If the total number of channels is less than or equal to MAX_CHANNELS_FOR_CALIBRATED_DISPLAY, all channels (both calibrated and uncalibrated) are included in their original order. - If the total number of channels exceeds this threshold, calibrated channels are excluded from the output, and all uncalibrated channels are shown in their original order.

channel_calibration_scores Computes per-channel calibration scores.
channels_recommended_for_calibration Returns uncalibrated channels with calibration score below threshold.
checks_status A dictionary of check names and statuses.
has_calibration_warning Returns True if any uncalibrated channel has score below threshold.

Methods

output_model_health_card

View source

output_model_health_card(
    filename: str, filepath: str
)

Generates and saves the HTML output for the model health card.

Args
filename The name of the file to save the HTML output to.
filepath The path to the directory to save the HTML output to.

__eq__

__eq__(
    other
)

Return self==value.

calibration_overview_data ()
high_variance_chart_json None
implausible_roi_chart_json None
potential_bias_chart_json None