Vmeasurescore

vmeasurescore

Metric VMeasureScore from torchmetrics (torchmetrics.clustering.VMeasureScore)

When to invoke this skill

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

Reference signature

from torchmetrics.clustering import VMeasureScore

# VMeasureScore(beta: float = 1.0, **kwargs: Any) -> None

Library docstring

Compute `V-Measure Score`_.

The V-measure is the harmonic mean between homogeneity and completeness:

.. math::
    v = \frac{(1 + \beta) * homogeneity * completeness}{\beta * homogeneity + completeness}

where :math:`\beta` is a weight parameter that defines the weight of homogeneity in the harmonic mean, with the
default value :math:`\beta=1`. The V-measure is symmetric, which means that swapping ``preds`` and ``target`` does
not change the score.

This clustering metric is an extrinsic measure, because it requires ground truth clustering labels, which may not
be available in practice since clustering in generally is 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:

- ``rand_score`` (:class:`~torch.Tensor`): A tensor with the Rand Score

Args:
    beta: Weight parameter that defines the weight of homogeneity in the harmonic mean
    kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info.

Example::
    >>> import torch
    >>> from torchmetrics.clustering import VMeasureScore
    >>> preds = torch.tensor([2, 1, 0, 1, 0])
    >>> target = torch.tensor([0, 2, 1, 1, 0])
    >>> metric = VMeasureScore(beta=2.0)
    >>> metric(preds, target)
    tensor(0.4744)

Quick recipe

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