pinball_loss#

openstef_beam.metrics.metrics_deterministic.pinball_loss(y_true: ndarray[tuple[Any, ...], dtype[floating]], y_pred: ndarray[tuple[Any, ...], dtype[floating]], *, quantile: Quantile, sample_weights: ndarray[tuple[Any, ...], dtype[floating]] | None = None) → float[source]#

Calculate the Pinball Loss (also known as Quantile Loss) for a single quantile.

The Pinball Loss is a scoring rule for a single quantile forecast that penalizes under- and over-predictions differently based on the quantile level.

Parameters:

  • y_true (ndarray[tuple[Any, ...], dtype[floating]]) – Ground truth values with shape (num_samples,).

  • y_pred (ndarray[tuple[Any, ...], dtype[floating]]) – Predicted quantile values with shape (num_samples,).

  • quantile (Quantile) – The quantile level being predicted (e.g., 0.1, 0.5, 0.9). Must be in [0, 1].

  • sample_weights (ndarray[tuple[Any, ...], dtype[floating]] | None) – Optional weights for each sample with shape (num_samples,). If None, all samples are weighted equally.

Returns:

The average Pinball Loss as a float. Lower is better.

Return type:

float

Example

Basic usage for 10th percentile predictions

>>> import numpy as np
>>> y_true = np.array([100, 120, 110])
>>> y_pred = np.array([95, 116, 111])
>>> pinball_loss(y_true, y_pred, quantile=0.1)
0.6
Parameters:
Return type:

float