MultiConfidenceMetrics(mapping=None, *, ignore_unknown_fields=False, **kwargs)
Metrics across multiple confidence levels.
Attributes | |
---|---|
Name | Description |
confidence_level_metrics |
MutableSequence[google.cloud.documentai_v1beta3.types.Evaluation.ConfidenceLevelMetrics]
Metrics across confidence levels with fuzzy matching enabled. |
confidence_level_metrics_exact |
MutableSequence[google.cloud.documentai_v1beta3.types.Evaluation.ConfidenceLevelMetrics]
Metrics across confidence levels with only exact matching. |
auprc |
float
The calculated area under the precision recall curve (AUPRC), computed by integrating over all confidence thresholds. |
estimated_calibration_error |
float
The Estimated Calibration Error (ECE) of the confidence of the predicted entities. |
auprc_exact |
float
The AUPRC for metrics with fuzzy matching disabled, i.e., exact matching only. |
estimated_calibration_error_exact |
float
The ECE for the predicted entities with fuzzy matching disabled, i.e., exact matching only. |
metrics_type |
google.cloud.documentai_v1beta3.types.Evaluation.MultiConfidenceMetrics.MetricsType
The metrics type for the label. |
Classes
MetricsType
MetricsType(value)
A type that determines how metrics should be interpreted.
Values: METRICS_TYPE_UNSPECIFIED (0): The metrics type is unspecified. By default, metrics without a particular specification are for leaf entity types (i.e., top-level entity types without child types, or child types which are not parent types themselves). AGGREGATE (1): Indicates whether metrics for this particular label type represent an aggregate of metrics for other types instead of being based on actual TP/FP/FN values for the label type. Metrics for parent (i.e., non-leaf) entity types are an aggregate of metrics for their children.