nicetoolbox.evaluation.metrics.categorical.pr_curve

Classes

PrCurveMetric

Precision-recall curve and average precision (AP) per pre-compute group.
class nicetoolbox.evaluation.metrics.categorical.pr_curve.PrCurveMetric(metrics_config: BaseMetricConfig, config_handler: ConfigHandler)[source]

Precision-recall curve and average precision (AP) per pre-compute group.

Predictions must be float confidence scores; ground truth must be bool. AP is computed once per group by pooling all frames — same rationale as RocAucMetric (avoids averaging per-pair scores across groups with different support and class balance).

Parameters:

  • metrics_config – Metric-specific configuration.

  • config_handler – Shared config handler with project and evaluation configs.

compute() → MetricResult[source]

Execute the metric end-to-end: load data, compute, return results.

Returns:

MetricResult containing summary tables, per-frame arrays, and plots.