PrecisionRecallCurvePlotter#

class openstef_beam.analysis.plots.PrecisionRecallCurvePlotter[source]

Bases: object

Creates precision-recall curves for model performance across quantiles.

This plotter evaluates forecast performance at different probability thresholds by visualizing the trade-off between precision and recall for each model.

The plots help answer:

  • Which model provides the best precision-recall trade-off?

  • How does performance vary across different quantile thresholds?

  • What quantile levels achieve optimal model performance?

The resulting curves show:

  • Precision vs recall curves for each model

  • Interactive hover data showing quantile values

  • Comparative model performance visualization

Higher curves (closer to top-right) indicate better overall performance, while specific points help identify optimal operating thresholds.

Example

Comparing model performance across quantiles

>>> from openstef_core.types import Quantile
>>> plotter = PrecisionRecallCurvePlotter()
>>>
>>> # Add model performance data
>>> precision = [0.8, 0.75, 0.7, 0.65]
>>> recall = [0.6, 0.7, 0.8, 0.85]
>>> quantiles = [Quantile(0.1), Quantile(0.3), Quantile(0.5), Quantile(0.9)]
>>> _ = plotter.add_model("XGBoost", precision, recall, quantiles)
>>> # Generate precision-recall curve
>>> fig = plotter.plot(title="Model Performance Analysis")
>>> type(fig).__name__
'Figure'
__init__()[source]

Initialize the plotter with empty model data storage.

add_model(model_name: str, precision_values: list[float], recall_values: list[float], quantiles: list[Quantile]) → PrecisionRecallCurvePlotter[source]

Add a model’s precision and recall values to the plot.

Parameters:

  • model_name (str) – The name of the model.

  • precision_values (list[float]) – List of precision values for each quantile.

  • recall_values (list[float]) – List of recall values for each quantile.

  • quantiles (list[Quantile]) – List of quantile values. If None, will use indices.

  • model_name

  • precision_values

  • recall_values

  • quantiles

Returns:

The current instance for method chaining.

Return type:

PrecisionRecallCurvePlotter

Raises:

ValueError – If precision and recall lists have different lengths, or if quantiles list has different length than precision/recall lists.

Return type:

PrecisionRecallCurvePlotter

plot(title: str = 'Precision-Recall Curve') → Figure[source]

Create and return a precision-recall curve plot for all added models.

Parameters:

  • title (str) – Title of the plot.

  • title

Returns:

The precision-recall curve plot.

Return type:

plotly.graph_objects.Figure

Raises:

ValueError – If no model data has been added.

Return type:

Figure