Adjustedrandscore

adjustedrandscore

Metric AdjustedRandScore from torchmetrics (torchmetrics.clustering.AdjustedRandScore)

When to invoke this skill

The user has predictions + ground truth and asks to evaluate with AdjustedRandScore, or mentions torchmetrics.clustering.AdjustedRandScore directly, or wants the standard torchmetrics implementation.

Reference signature

from torchmetrics.clustering import AdjustedRandScore

# AdjustedRandScore(**kwargs: Any) -> None

Library docstring

Compute `Adjusted Rand Score`_ (also known as Adjusted Rand Index).

.. math::
    ARS(U, V) = (\text{RS} - \text{Expected RS}) / (\text{Max RS} - \text{Expected RS})

The adjusted rand score :math:`\text{ARS}` is in essence the :math:`\text{RS}` (rand score) adjusted for chance.
The score ensures that completely randomly cluster labels have a score close to zero and only a perfect match will
have a score of 1 (up to a permutation of the labels). The adjusted rand score is symmetric, therefore swapping
:math:`U` and :math:`V` yields the same adjusted rand score.

This clustering metric is an extrinsic measure, because it requires ground truth clustering labels, which may not
be available in practice since clustering is generally used for unsupervised learning.

As input to ``forward`` and ``update`` the metric accepts the following input:

- ``preds`` (:class:`~torch.Tensor`): single integer tensor with shape ``(N,)`` with predicted cluster labels
- ``target`` (:class:`~torch.Tensor`): single integer tensor with shape ``(N,)`` with ground truth cluster labels

As output of ``forward`` and ``compute`` the metric returns the following output:

- ``adj_rand_score`` (:class:`~torch.Tensor`): Scalar tensor with the adjusted rand score

Args:
    kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info.

Example::
    >>> import torch
    >>> from torchmetrics.clustering import AdjustedRandScore
    >>> metric = AdjustedRandScore()
    >>> metric(torch.tensor([0, 0, 1, 1]), torch.tensor([0, 0, 1, 1]))
    tensor(1.)
    >>> metric(torch.tensor([0, 0, 1, 1]), torch.tensor([0, 1, 0, 1]))
    tensor(-0.5000)

Quick recipe

import torchmetrics.clustering as _m
score = _m.AdjustedRandScore(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).