mean_pinball_loss#

openstef_beam.metrics.metrics_probabilistic.mean_pinball_loss(y_true: ndarray[tuple[Any, ...], dtype[floating]], y_pred: ndarray[tuple[Any, ...], dtype[floating]], quantiles: list[Quantile], sample_weight: ndarray[tuple[Any, ...], dtype[floating]] | None = None) → float[source]#

Calculate the Mean Pinball Loss for quantile forecasts.

The Pinball Loss is a proper scoring rule for evaluating quantile forecasts. It penalizes under- and over-predictions differently based on the quantile level.

Parameters:

  • y_true (ndarray[tuple[Any, ...], dtype[floating]]) – Observed values with shape (num_samples,) or (num_samples, num_quantiles).

  • y_pred (ndarray[tuple[Any, ...], dtype[floating]]) – Predicted quantiles with shape (num_samples, num_quantiles). Each column corresponds to predictions for a specific quantile level.

  • quantiles (list[Quantile]) – Quantile levels with shape (num_quantiles,). Must be sorted in ascending order and contain values in [0, 1].

  • sample_weight (ndarray[tuple[Any, ...], dtype[floating]] | None) – Optional weights for each sample with shape (num_samples,).

  • y_true

  • y_pred

  • quantiles

  • sample_weight

Returns:

The weighted average Pinball Loss across all samples and quantiles. Lower values indicate better forecast quality.

Return type:

float