torcheval.metrics.functional.binary_recall_at_fixed_precision

torcheval.metrics.functional.binary_recall_at_fixed_precision(input: Tensor, target: Tensor, *, min_precision: float) → Tuple[Tensor, Tensor]

Returns the highest possible recall value given the minimum precision for binary classification tasks.

Its class version is torcheval.metrics.BinaryRecallAtFixedPrecision. See also multilabel_recall_at_fixed_precision

Parameters:

• input (Tensor) – Tensor of label predictions It should be probabilities with shape of (n_samples, )
• target (Tensor) – Tensor of ground truth labels with shape of (n_samples, )
• min_precision (float) – Minimum precision threshold

Returns:

• recall (Tensor): Max recall value given the minimum precision
• thresholds (Tensor): Corresponding threshold to max recall

Return type:

Tuple

Examples:

>>> import torch
>>> from torcheval.metrics.functional import binary_recall_at_fixed_precision
>>> input = torch.tensor([0.1, 0.4, 0.6, 0.6, 0.6, 0.35, 0.8])
>>> target = torch.tensor([0, 0, 1, 1, 1, 1, 1])
>>> binary_recall_at_fixed_precision(input, target, min_precision=0.5)
(torch.tensor(1.0), torch.tensor(0.35))