Mean Pinball Loss

mean-pinball-loss

Metric mean_pinball_loss from scikit-learn (sklearn.metrics.mean_pinball_loss)

When to invoke this skill

The user has predictions + ground truth and asks to evaluate with mean_pinball_loss, or mentions sklearn.metrics.mean_pinball_loss directly, or wants the standard scikit-learn implementation.

Reference signature

from sklearn.metrics import mean_pinball_loss

# mean_pinball_loss(y_true, y_pred, *, sample_weight=None, alpha=0.5, multioutput='uniform_average')

Library docstring

Pinball loss for quantile regression.

Read more in the :ref:`User Guide <pinball_loss>`.

Parameters
----------
y_true : array-like of shape (n_samples,) or (n_samples, n_outputs)
    Ground truth (correct) target values.

y_pred : array-like of shape (n_samples,) or (n_samples, n_outputs)
    Estimated target values.

sample_weight : array-like of shape (n_samples,), default=None
    Sample weights.

alpha : float, slope of the pinball loss, default=0.5,
    This loss is equivalent to :ref:`mean_absolute_error` when `alpha=0.5`,
    `alpha=0.95` is minimized by estimators of the 95th percentile.

multioutput : {'raw_values', 'uniform_average'}  or array-like of shape             (n_outputs,), default='uniform_average'
    Defines aggregating of multiple output values.
    Array-like value defines weights used to average errors.

    'raw_values' :
        Returns a full set of errors in case of multioutput input.

    'uniform_average' :
        Errors of all outputs are averaged with uniform weight.

Returns
-------
loss : float or ndarray of floats
    If multioutput is 'raw_values', then mean absolute error is returned
    for each output separately.
    If multioutput is 'uniform_average' or an ndarray of weights, then the
    weighted average of all output errors is returned.

    The pinball loss output is a non-negative floating point. The best
    value is 0.0.

Examples
--------
>>> from sklearn.metrics import mean_pinball_loss
>>> y_true = [1, 2, 3]
>>> mean_pinball_loss(y_true, [0, 2, 3], alpha=0.1)
0.03...
>>> mean_pinball_loss(y_true, [1, 2, 4], alpha=0.1)
0.3...
>>> mean_pinball_loss(y_true, [0, 2, 3], alpha=0.9)
0.3...
>>> mean_pinball_loss(y_true, [1, 2, 4], alpha=0.9)
0.03...
>>> mean_pinball_loss(y_true, y_true, alpha=0.1)
0.0
>>> mean_pinball_loss(y_true, y_true, alpha=0.9)
0.0

Quick recipe

import sklearn.metrics as _m
score = _m.mean_pinball_loss(y_true, y_pred)

Don'ts

• Don't reimplement when the library version handles edge cases (NaN, ties, empty inputs) better than a hand-rolled formula.
• Always check the library version's argument order — sklearn is (y_true, y_pred) while torchmetrics is (preds, target).