probly.plot.ood.plot_roc_curve

probly.plot.ood.plot_roc_curve(fpr: np.ndarray, tpr: np.ndarray, auroc: float, fpr95: float | None = None, ax: Axes | None = None, config: PlotConfig | None = None) → Figure | SubFigure[source]

Plot a Receiver Operating Characteristic (ROC) curve.

The curve is annotated with the AUROC score and, optionally, the false positive rate at 95 % TPR. A diagonal reference line representing a random classifier is included for visual calibration.

Parameters:

  • fpr – False positive rate values as returned by sklearn.metrics.roc_curve.

  • tpr – True positive rate values as returned by sklearn.metrics.roc_curve.

  • auroc – Area under the ROC curve.

  • fpr95 – False positive rate at 95 % TPR. When provided it is appended to the legend label.

  • ax – Optional pre-existing axes for subplot composition. When None a new figure is created.

  • config – Plot configuration. Defaults to PlotConfig() when None.

Returns:

The figure that contains the ROC curve.

Raises:

RuntimeError – If ax is provided but has no associated figure.