Precision Recall Curve¶

Module Interface¶

class torchmetrics.PrecisionRecallCurve(task: Literal['binary', 'multiclass', 'multilabel'], thresholds: Optional[Union[int, List[float], torch.Tensor]] = None, num_classes: Optional[int] = None, num_labels: Optional[int] = None, ignore_index: Optional[int] = None, validate_args: bool = True, **kwargs: Any)[source]

Computes the precision-recall curve. The curve consist of multiple pairs of precision and recall values evaluated at different thresholds, such that the tradeoff between the two values can been seen.

This function is a simple wrapper to get the task specific versions of this metric, which is done by setting the task argument to either 'binary', 'multiclass' or multilabel. See the documentation of BinaryPrecisionRecallCurve, MulticlassPrecisionRecallCurve and MultilabelPrecisionRecallCurve for the specific details of each argument influence and examples.

Legacy Example:

>>> pred = torch.tensor([0, 0.1, 0.8, 0.4])
>>> target = torch.tensor([0, 1, 1, 0])
>>> pr_curve = PrecisionRecallCurve(task="binary")
>>> precision, recall, thresholds = pr_curve(pred, target)
>>> precision
tensor([0.6667, 0.5000, 1.0000, 1.0000])
>>> recall
tensor([1.0000, 0.5000, 0.5000, 0.0000])
>>> thresholds
tensor([0.1000, 0.4000, 0.8000])

>>> pred = torch.tensor([[0.75, 0.05, 0.05, 0.05, 0.05],
...                      [0.05, 0.75, 0.05, 0.05, 0.05],
...                      [0.05, 0.05, 0.75, 0.05, 0.05],
...                      [0.05, 0.05, 0.05, 0.75, 0.05]])
>>> target = torch.tensor([0, 1, 3, 2])
>>> pr_curve = PrecisionRecallCurve(task="multiclass", num_classes=5)
>>> precision, recall, thresholds = pr_curve(pred, target)
>>> precision
[tensor([1., 1.]), tensor([1., 1.]), tensor([0.2500, 0.0000, 1.0000]),
 tensor([0.2500, 0.0000, 1.0000]), tensor([0., 1.])]
>>> recall
[tensor([1., 0.]), tensor([1., 0.]), tensor([1., 0., 0.]), tensor([1., 0., 0.]), tensor([nan, 0.])]
>>> thresholds
[tensor(0.7500), tensor(0.7500), tensor([0.0500, 0.7500]), tensor([0.0500, 0.7500]), tensor(0.0500)]

BinaryPrecisionRecallCurve¶

class torchmetrics.classification.BinaryPrecisionRecallCurve(thresholds=None, ignore_index=None, validate_args=True, **kwargs)[source]

Computes the precision-recall curve for binary tasks. The curve consist of multiple pairs of precision and recall values evaluated at different thresholds, such that the tradeoff between the two values can been seen.

Accepts the following input tensors:

preds (float tensor): (N, ...). Preds should be a tensor containing probabilities or logits for each observation. If preds has values outside [0,1] range we consider the input to be logits and will auto apply sigmoid per element.
target (int tensor): (N, ...). Target should be a tensor containing ground truth labels, and therefore only contain {0,1} values (except if ignore_index is specified).

Additional dimension ... will be flattened into the batch dimension.

The implementation both supports calculating the metric in a non-binned but accurate version and a binned version that is less accurate but more memory efficient. Setting the thresholds argument to None will activate the non-binned version that uses memory of size $\mathcal{O}(n_{samples})$ whereas setting the thresholds argument to either an integer, list or a 1d tensor will use a binned version that uses memory of size $\mathcal{O}(n_{thresholds})$ (constant memory).

Parameters

thresholds¶ (Union[int, List[float], Tensor, None]) –
Can be one of:
- If set to None, will use a non-binned approach where thresholds are dynamically calculated from all the data. Most accurate but also most memory consuming approach.
- If set to an int (larger than 1), will use that number of thresholds linearly spaced from 0 to 1 as bins for the calculation.
- If set to an list of floats, will use the indicated thresholds in the list as bins for the calculation
- If set to an 1d tensor of floats, will use the indicated thresholds in the tensor as bins for the calculation.
validate_args¶ (bool) – bool indicating if input arguments and tensors should be validated for correctness. Set to False for faster computations.
kwargs¶ (Any) – Additional keyword arguments, see Advanced metric settings for more info.

Returns

a tuple of 3 tensors containing:

precision: an 1d tensor of size (n_thresholds+1, ) with precision values
recall: an 1d tensor of size (n_thresholds+1, ) with recall values
thresholds: an 1d tensor of size (n_thresholds, ) with increasing threshold values

Return type

(tuple)

Example

>>> from torchmetrics.classification import BinaryPrecisionRecallCurve
>>> preds = torch.tensor([0, 0.5, 0.7, 0.8])
>>> target = torch.tensor([0, 1, 1, 0])
>>> metric = BinaryPrecisionRecallCurve(thresholds=None)
>>> metric(preds, target)  
(tensor([0.6667, 0.5000, 0.0000, 1.0000]),
 tensor([1.0000, 0.5000, 0.0000, 0.0000]),
 tensor([0.5000, 0.7000, 0.8000]))
>>> metric = BinaryPrecisionRecallCurve(thresholds=5)
>>> metric(preds, target)  
(tensor([0.5000, 0.6667, 0.6667, 0.0000, 0.0000, 1.0000]),
 tensor([1., 1., 1., 0., 0., 0.]),
 tensor([0.0000, 0.2500, 0.5000, 0.7500, 1.0000]))

Initializes internal Module state, shared by both nn.Module and ScriptModule.

compute()[source]

Override this method to compute the final metric value from state variables synchronized across the distributed backend.

Return type: Tuple[Tensor, Tensor, Tensor]

update(preds, target)[source]

Override this method to update the state variables of your metric class.

Return type: None

MulticlassPrecisionRecallCurve¶

class torchmetrics.classification.MulticlassPrecisionRecallCurve(num_classes, thresholds=None, ignore_index=None, validate_args=True, **kwargs)[source]

Computes the precision-recall curve for multiclass tasks. The curve consist of multiple pairs of precision and recall values evaluated at different thresholds, such that the tradeoff between the two values can been seen.

Accepts the following input tensors:

preds (float tensor): (N, C, ...). Preds should be a tensor containing probabilities or logits for each observation. If preds has values outside [0,1] range we consider the input to be logits and will auto apply softmax per sample.
target (int tensor): (N, ...). Target should be a tensor containing ground truth labels, and therefore only contain values in the [0, n_classes-1] range (except if ignore_index is specified).

Additional dimension ... will be flattened into the batch dimension.

The implementation both supports calculating the metric in a non-binned but accurate version and a binned version that is less accurate but more memory efficient. Setting the thresholds argument to None will activate the non-binned version that uses memory of size $\mathcal{O}(n_{samples})$ whereas setting the thresholds argument to either an integer, list or a 1d tensor will use a binned version that uses memory of size $\mathcal{O}(n_{thresholds} \times n_{classes})$ (constant memory).

Parameters

num_classes¶ (int) – Integer specifing the number of classes
thresholds¶ (Union[int, List[float], Tensor, None]) –
Can be one of:
- If set to None, will use a non-binned approach where thresholds are dynamically calculated from all the data. Most accurate but also most memory consuming approach.
- If set to an int (larger than 1), will use that number of thresholds linearly spaced from 0 to 1 as bins for the calculation.
- If set to an list of floats, will use the indicated thresholds in the list as bins for the calculation
- If set to an 1d tensor of floats, will use the indicated thresholds in the tensor as bins for the calculation.
validate_args¶ (bool) – bool indicating if input arguments and tensors should be validated for correctness. Set to False for faster computations.
kwargs¶ (Any) – Additional keyword arguments, see Advanced metric settings for more info.

Returns

a tuple of either 3 tensors or 3 lists containing

precision: if thresholds=None a list for each class is returned with an 1d tensor of size (n_thresholds+1, ) with precision values (length may differ between classes). If thresholds is set to something else, then a single 2d tensor of size (n_classes, n_thresholds+1) with precision values is returned.
recall: if thresholds=None a list for each class is returned with an 1d tensor of size (n_thresholds+1, ) with recall values (length may differ between classes). If thresholds is set to something else, then a single 2d tensor of size (n_classes, n_thresholds+1) with recall values is returned.
thresholds: if thresholds=None a list for each class is returned with an 1d tensor of size (n_thresholds, ) with increasing threshold values (length may differ between classes). If threshold is set to something else, then a single 1d tensor of size (n_thresholds, ) is returned with shared threshold values for all classes.

Return type

(tuple)

Example

>>> from torchmetrics.classification import MulticlassPrecisionRecallCurve
>>> preds = torch.tensor([[0.75, 0.05, 0.05, 0.05, 0.05],
...                       [0.05, 0.75, 0.05, 0.05, 0.05],
...                       [0.05, 0.05, 0.75, 0.05, 0.05],
...                       [0.05, 0.05, 0.05, 0.75, 0.05]])
>>> target = torch.tensor([0, 1, 3, 2])
>>> metric = MulticlassPrecisionRecallCurve(num_classes=5, thresholds=None)
>>> precision, recall, thresholds = metric(preds, target)
>>> precision  
[tensor([1., 1.]), tensor([1., 1.]), tensor([0.2500, 0.0000, 1.0000]),
 tensor([0.2500, 0.0000, 1.0000]), tensor([0., 1.])]
>>> recall
[tensor([1., 0.]), tensor([1., 0.]), tensor([1., 0., 0.]), tensor([1., 0., 0.]), tensor([nan, 0.])]
>>> thresholds
[tensor(0.7500), tensor(0.7500), tensor([0.0500, 0.7500]), tensor([0.0500, 0.7500]), tensor(0.0500)]
>>> metric = MulticlassPrecisionRecallCurve(num_classes=5, thresholds=5)
>>> metric(preds, target)  
(tensor([[0.2500, 1.0000, 1.0000, 1.0000, 0.0000, 1.0000],
         [0.2500, 1.0000, 1.0000, 1.0000, 0.0000, 1.0000],
         [0.2500, 0.0000, 0.0000, 0.0000, 0.0000, 1.0000],
         [0.2500, 0.0000, 0.0000, 0.0000, 0.0000, 1.0000],
         [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 1.0000]]),
 tensor([[1., 1., 1., 1., 0., 0.],
         [1., 1., 1., 1., 0., 0.],
         [1., 0., 0., 0., 0., 0.],
         [1., 0., 0., 0., 0., 0.],
         [0., 0., 0., 0., 0., 0.]]),
 tensor([0.0000, 0.2500, 0.5000, 0.7500, 1.0000]))

Initializes internal Module state, shared by both nn.Module and ScriptModule.

compute()[source]

Override this method to compute the final metric value from state variables synchronized across the distributed backend.

Return type: Union[Tuple[Tensor, Tensor, Tensor], Tuple[List[Tensor], List[Tensor], List[Tensor]]]

update(preds, target)[source]

Override this method to update the state variables of your metric class.

Return type: None

MultilabelPrecisionRecallCurve¶

class torchmetrics.classification.MultilabelPrecisionRecallCurve(num_labels, thresholds=None, ignore_index=None, validate_args=True, **kwargs)[source]

Computes the precision-recall curve for multilabel tasks. The curve consist of multiple pairs of precision and recall values evaluated at different thresholds, such that the tradeoff between the two values can been seen.

Accepts the following input tensors:

preds (float tensor): (N, C, ...). Preds should be a tensor containing probabilities or logits for each observation. If preds has values outside [0,1] range we consider the input to be logits and will auto apply sigmoid per element.
target (int tensor): (N, C, ...). Target should be a tensor containing ground truth labels, and therefore only contain {0,1} values (except if ignore_index is specified).

Additional dimension ... will be flattened into the batch dimension.

The implementation both supports calculating the metric in a non-binned but accurate version and a binned version that is less accurate but more memory efficient. Setting the thresholds argument to None will activate the non-binned version that uses memory of size $\mathcal{O}(n_{samples})$ whereas setting the thresholds argument to either an integer, list or a 1d tensor will use a binned version that uses memory of size $\mathcal{O}(n_{thresholds} \times n_{labels})$ (constant memory).

Parameters

preds¶ – Tensor with predictions
target¶ – Tensor with true labels
num_labels¶ (int) – Integer specifing the number of labels
thresholds¶ (Union[int, List[float], Tensor, None]) –
Can be one of:
- If set to None, will use a non-binned approach where thresholds are dynamically calculated from all the data. Most accurate but also most memory consuming approach.
- If set to an int (larger than 1), will use that number of thresholds linearly spaced from 0 to 1 as bins for the calculation.
- If set to an list of floats, will use the indicated thresholds in the list as bins for the calculation
- If set to an 1d tensor of floats, will use the indicated thresholds in the tensor as bins for the calculation.
validate_args¶ (bool) – bool indicating if input arguments and tensors should be validated for correctness. Set to False for faster computations.

Returns

a tuple of either 3 tensors or 3 lists containing

precision: if thresholds=None a list for each label is returned with an 1d tensor of size (n_thresholds+1, ) with precision values (length may differ between labels). If thresholds is set to something else, then a single 2d tensor of size (n_labels, n_thresholds+1) with precision values is returned.
recall: if thresholds=None a list for each label is returned with an 1d tensor of size (n_thresholds+1, ) with recall values (length may differ between labels). If thresholds is set to something else, then a single 2d tensor of size (n_labels, n_thresholds+1) with recall values is returned.
thresholds: if thresholds=None a list for each label is returned with an 1d tensor of size (n_thresholds, ) with increasing threshold values (length may differ between labels). If threshold is set to something else, then a single 1d tensor of size (n_thresholds, ) is returned with shared threshold values for all labels.

Return type

(tuple)

Example

>>> from torchmetrics.classification import MultilabelPrecisionRecallCurve
>>> preds = torch.tensor([[0.75, 0.05, 0.35],
...                       [0.45, 0.75, 0.05],
...                       [0.05, 0.55, 0.75],
...                       [0.05, 0.65, 0.05]])
>>> target = torch.tensor([[1, 0, 1],
...                        [0, 0, 0],
...                        [0, 1, 1],
...                        [1, 1, 1]])
>>> metric = MultilabelPrecisionRecallCurve(num_labels=3, thresholds=None)
>>> precision, recall, thresholds = metric(preds, target)
>>> precision  
[tensor([0.5000, 0.5000, 1.0000, 1.0000]), tensor([0.6667, 0.5000, 0.0000, 1.0000]),
 tensor([0.7500, 1.0000, 1.0000, 1.0000])]
>>> recall  
[tensor([1.0000, 0.5000, 0.5000, 0.0000]), tensor([1.0000, 0.5000, 0.0000, 0.0000]),
 tensor([1.0000, 0.6667, 0.3333, 0.0000])]
>>> thresholds  
[tensor([0.0500, 0.4500, 0.7500]), tensor([0.5500, 0.6500, 0.7500]),
 tensor([0.0500, 0.3500, 0.7500])]
>>> metric = MultilabelPrecisionRecallCurve(num_labels=3, thresholds=5)
>>> metric(preds, target)  
(tensor([[0.5000, 0.5000, 1.0000, 1.0000, 0.0000, 1.0000],
         [0.5000, 0.6667, 0.6667, 0.0000, 0.0000, 1.0000],
         [0.7500, 1.0000, 1.0000, 1.0000, 0.0000, 1.0000]]),
 tensor([[1.0000, 0.5000, 0.5000, 0.5000, 0.0000, 0.0000],
         [1.0000, 1.0000, 1.0000, 0.0000, 0.0000, 0.0000],
         [1.0000, 0.6667, 0.3333, 0.3333, 0.0000, 0.0000]]),
 tensor([0.0000, 0.2500, 0.5000, 0.7500, 1.0000]))

Initializes internal Module state, shared by both nn.Module and ScriptModule.

compute()[source]

Override this method to compute the final metric value from state variables synchronized across the distributed backend.

Return type: Union[Tuple[Tensor, Tensor, Tensor], Tuple[List[Tensor], List[Tensor], List[Tensor]]]

update(preds, target)[source]

Override this method to update the state variables of your metric class.

Return type: None

Functional Interface¶

torchmetrics.functional.precision_recall_curve(preds, target, task, thresholds=None, num_classes=None, num_labels=None, ignore_index=None, validate_args=True)[source]

Computes the precision-recall curve. The curve consist of multiple pairs of precision and recall values evaluated at different thresholds, such that the tradeoff between the two values can been seen.

This function is a simple wrapper to get the task specific versions of this metric, which is done by setting the task argument to either 'binary', 'multiclass' or multilabel. See the documentation of binary_precision_recall_curve(), multiclass_precision_recall_curve() and multilabel_precision_recall_curve() for the specific details of each argument influence and examples.

Legacy Example:

>>> pred = torch.tensor([0.0, 1.0, 2.0, 3.0])
>>> target = torch.tensor([0, 1, 1, 0])
>>> precision, recall, thresholds = precision_recall_curve(pred, target, task='binary')
>>> precision
tensor([0.6667, 0.5000, 0.0000, 1.0000])
>>> recall
tensor([1.0000, 0.5000, 0.0000, 0.0000])
>>> thresholds
tensor([0.7311, 0.8808, 0.9526])

>>> pred = torch.tensor([[0.75, 0.05, 0.05, 0.05, 0.05],
...                      [0.05, 0.75, 0.05, 0.05, 0.05],
...                      [0.05, 0.05, 0.75, 0.05, 0.05],
...                      [0.05, 0.05, 0.05, 0.75, 0.05]])
>>> target = torch.tensor([0, 1, 3, 2])
>>> precision, recall, thresholds = precision_recall_curve(pred, target, task='multiclass', num_classes=5)
>>> precision
[tensor([1., 1.]), tensor([1., 1.]), tensor([0.2500, 0.0000, 1.0000]),
 tensor([0.2500, 0.0000, 1.0000]), tensor([0., 1.])]
>>> recall
[tensor([1., 0.]), tensor([1., 0.]), tensor([1., 0., 0.]), tensor([1., 0., 0.]), tensor([nan, 0.])]
>>> thresholds
[tensor([0.7500]), tensor([0.7500]), tensor([0.0500, 0.7500]), tensor([0.0500, 0.7500]), tensor([0.0500])]

Return type: Union[Tuple[Tensor, Tensor, Tensor], Tuple[List[Tensor], List[Tensor], List[Tensor]]]

binary_precision_recall_curve¶

torchmetrics.functional.classification.binary_precision_recall_curve(preds, target, thresholds=None, ignore_index=None, validate_args=True)[source]

Computes the precision-recall curve for binary tasks. The curve consist of multiple pairs of precision and recall values evaluated at different thresholds, such that the tradeoff between the two values can been seen.

Accepts the following input tensors:

preds (float tensor): (N, ...). Preds should be a tensor containing probabilities or logits for each observation. If preds has values outside [0,1] range we consider the input to be logits and will auto apply sigmoid per element.
target (int tensor): (N, ...). Target should be a tensor containing ground truth labels, and therefore only contain {0,1} values (except if ignore_index is specified).

Additional dimension ... will be flattened into the batch dimension.

The implementation both supports calculating the metric in a non-binned but accurate version and a binned version that is less accurate but more memory efficient. Setting the thresholds argument to None will activate the non-binned version that uses memory of size $\mathcal{O}(n_{samples})$ whereas setting the thresholds argument to either an integer, list or a 1d tensor will use a binned version that uses memory of size $\mathcal{O}(n_{thresholds})$ (constant memory).

Parameters

preds¶ (Tensor) – Tensor with predictions
target¶ (Tensor) – Tensor with true labels
thresholds¶ (Union[int, List[float], Tensor, None]) –
Can be one of:
- If set to None, will use a non-binned approach where thresholds are dynamically calculated from all the data. Most accurate but also most memory consuming approach.
- If set to an int (larger than 1), will use that number of thresholds linearly spaced from 0 to 1 as bins for the calculation.
- If set to an list of floats, will use the indicated thresholds in the list as bins for the calculation
- If set to an 1d tensor of floats, will use the indicated thresholds in the tensor as bins for the calculation.
validate_args¶ (bool) – bool indicating if input arguments and tensors should be validated for correctness. Set to False for faster computations.

Returns

a tuple of 3 tensors containing:

precision: an 1d tensor of size (n_thresholds+1, ) with precision values
recall: an 1d tensor of size (n_thresholds+1, ) with recall values
thresholds: an 1d tensor of size (n_thresholds, ) with increasing threshold values

Return type

(tuple)

Example

>>> from torchmetrics.functional.classification import binary_precision_recall_curve
>>> preds = torch.tensor([0, 0.5, 0.7, 0.8])
>>> target = torch.tensor([0, 1, 1, 0])
>>> binary_precision_recall_curve(preds, target, thresholds=None)  
(tensor([0.6667, 0.5000, 0.0000, 1.0000]),
 tensor([1.0000, 0.5000, 0.0000, 0.0000]),
 tensor([0.5000, 0.7000, 0.8000]))
>>> binary_precision_recall_curve(preds, target, thresholds=5)  
(tensor([0.5000, 0.6667, 0.6667, 0.0000, 0.0000, 1.0000]),
 tensor([1., 1., 1., 0., 0., 0.]),
 tensor([0.0000, 0.2500, 0.5000, 0.7500, 1.0000]))

multiclass_precision_recall_curve¶

torchmetrics.functional.classification.multiclass_precision_recall_curve(preds, target, num_classes, thresholds=None, ignore_index=None, validate_args=True)[source]

Computes the precision-recall curve for multiclass tasks. The curve consist of multiple pairs of precision and recall values evaluated at different thresholds, such that the tradeoff between the two values can been seen.

Accepts the following input tensors:

preds (float tensor): (N, C, ...). Preds should be a tensor containing probabilities or logits for each observation. If preds has values outside [0,1] range we consider the input to be logits and will auto apply softmax per sample.
target (int tensor): (N, ...). Target should be a tensor containing ground truth labels, and therefore only contain values in the [0, n_classes-1] range (except if ignore_index is specified).

Additional dimension ... will be flattened into the batch dimension.

The implementation both supports calculating the metric in a non-binned but accurate version and a binned version that is less accurate but more memory efficient. Setting the thresholds argument to None will activate the non-binned version that uses memory of size $\mathcal{O}(n_{samples})$ whereas setting the thresholds argument to either an integer, list or a 1d tensor will use a binned version that uses memory of size $\mathcal{O}(n_{thresholds} \times n_{classes})$ (constant memory).

Parameters

preds¶ (Tensor) – Tensor with predictions
target¶ (Tensor) – Tensor with true labels
num_classes¶ (int) – Integer specifing the number of classes
thresholds¶ (Union[int, List[float], Tensor, None]) –
Can be one of:
- If set to None, will use a non-binned approach where thresholds are dynamically calculated from all the data. Most accurate but also most memory consuming approach.
- If set to an int (larger than 1), will use that number of thresholds linearly spaced from 0 to 1 as bins for the calculation.
- If set to an list of floats, will use the indicated thresholds in the list as bins for the calculation
- If set to an 1d tensor of floats, will use the indicated thresholds in the tensor as bins for the calculation.
validate_args¶ (bool) – bool indicating if input arguments and tensors should be validated for correctness. Set to False for faster computations.

Returns

a tuple of either 3 tensors or 3 lists containing

precision: if thresholds=None a list for each class is returned with an 1d tensor of size (n_thresholds+1, ) with precision values (length may differ between classes). If thresholds is set to something else, then a single 2d tensor of size (n_classes, n_thresholds+1) with precision values is returned.
recall: if thresholds=None a list for each class is returned with an 1d tensor of size (n_thresholds+1, ) with recall values (length may differ between classes). If thresholds is set to something else, then a single 2d tensor of size (n_classes, n_thresholds+1) with recall values is returned.
thresholds: if thresholds=None a list for each class is returned with an 1d tensor of size (n_thresholds, ) with increasing threshold values (length may differ between classes). If threshold is set to something else, then a single 1d tensor of size (n_thresholds, ) is returned with shared threshold values for all classes.

Return type

(tuple)

Example

>>> from torchmetrics.functional.classification import multiclass_precision_recall_curve
>>> preds = torch.tensor([[0.75, 0.05, 0.05, 0.05, 0.05],
...                       [0.05, 0.75, 0.05, 0.05, 0.05],
...                       [0.05, 0.05, 0.75, 0.05, 0.05],
...                       [0.05, 0.05, 0.05, 0.75, 0.05]])
>>> target = torch.tensor([0, 1, 3, 2])
>>> precision, recall, thresholds = multiclass_precision_recall_curve(
...    preds, target, num_classes=5, thresholds=None
... )
>>> precision  
[tensor([1., 1.]), tensor([1., 1.]), tensor([0.2500, 0.0000, 1.0000]),
 tensor([0.2500, 0.0000, 1.0000]), tensor([0., 1.])]
>>> recall
[tensor([1., 0.]), tensor([1., 0.]), tensor([1., 0., 0.]), tensor([1., 0., 0.]), tensor([nan, 0.])]
>>> thresholds
[tensor([0.7500]), tensor([0.7500]), tensor([0.0500, 0.7500]), tensor([0.0500, 0.7500]), tensor([0.0500])]
>>> multiclass_precision_recall_curve(
...     preds, target, num_classes=5, thresholds=5
... )  
(tensor([[0.2500, 1.0000, 1.0000, 1.0000, 0.0000, 1.0000],
         [0.2500, 1.0000, 1.0000, 1.0000, 0.0000, 1.0000],
         [0.2500, 0.0000, 0.0000, 0.0000, 0.0000, 1.0000],
         [0.2500, 0.0000, 0.0000, 0.0000, 0.0000, 1.0000],
         [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 1.0000]]),
 tensor([[1., 1., 1., 1., 0., 0.],
         [1., 1., 1., 1., 0., 0.],
         [1., 0., 0., 0., 0., 0.],
         [1., 0., 0., 0., 0., 0.],
         [0., 0., 0., 0., 0., 0.]]),
 tensor([0.0000, 0.2500, 0.5000, 0.7500, 1.0000]))

multilabel_precision_recall_curve¶

torchmetrics.functional.classification.multilabel_precision_recall_curve(preds, target, num_labels, thresholds=None, ignore_index=None, validate_args=True)[source]

Computes the precision-recall curve for multilabel tasks. The curve consist of multiple pairs of precision and recall values evaluated at different thresholds, such that the tradeoff between the two values can been seen.

Accepts the following input tensors:

preds (float tensor): (N, C, ...). Preds should be a tensor containing probabilities or logits for each observation. If preds has values outside [0,1] range we consider the input to be logits and will auto apply sigmoid per element.
target (int tensor): (N, C, ...). Target should be a tensor containing ground truth labels, and therefore only contain {0,1} values (except if ignore_index is specified).

Additional dimension ... will be flattened into the batch dimension.

The implementation both supports calculating the metric in a non-binned but accurate version and a binned version that is less accurate but more memory efficient. Setting the thresholds argument to None will activate the non-binned version that uses memory of size $\mathcal{O}(n_{samples})$ whereas setting the thresholds argument to either an integer, list or a 1d tensor will use a binned version that uses memory of size $\mathcal{O}(n_{thresholds} \times n_{labels})$ (constant memory).

Parameters

preds¶ (Tensor) – Tensor with predictions
target¶ (Tensor) – Tensor with true labels
num_labels¶ (int) – Integer specifing the number of labels
thresholds¶ (Union[int, List[float], Tensor, None]) –
Can be one of:
- If set to None, will use a non-binned approach where thresholds are dynamically calculated from all the data. Most accurate but also most memory consuming approach.
- If set to an int (larger than 1), will use that number of thresholds linearly spaced from 0 to 1 as bins for the calculation.
- If set to an list of floats, will use the indicated thresholds in the list as bins for the calculation
- If set to an 1d tensor of floats, will use the indicated thresholds in the tensor as bins for the calculation.
validate_args¶ (bool) – bool indicating if input arguments and tensors should be validated for correctness. Set to False for faster computations.

Returns

a tuple of either 3 tensors or 3 lists containing

precision: if thresholds=None a list for each label is returned with an 1d tensor of size (n_thresholds+1, ) with precision values (length may differ between labels). If thresholds is set to something else, then a single 2d tensor of size (n_labels, n_thresholds+1) with precision values is returned.
recall: if thresholds=None a list for each label is returned with an 1d tensor of size (n_thresholds+1, ) with recall values (length may differ between labels). If thresholds is set to something else, then a single 2d tensor of size (n_labels, n_thresholds+1) with recall values is returned.
thresholds: if thresholds=None a list for each label is returned with an 1d tensor of size (n_thresholds, ) with increasing threshold values (length may differ between labels). If threshold is set to something else, then a single 1d tensor of size (n_thresholds, ) is returned with shared threshold values for all labels.

Return type

(tuple)

Example

>>> from torchmetrics.functional.classification import multilabel_precision_recall_curve
>>> preds = torch.tensor([[0.75, 0.05, 0.35],
...                       [0.45, 0.75, 0.05],
...                       [0.05, 0.55, 0.75],
...                       [0.05, 0.65, 0.05]])
>>> target = torch.tensor([[1, 0, 1],
...                        [0, 0, 0],
...                        [0, 1, 1],
...                        [1, 1, 1]])
>>> precision, recall, thresholds = multilabel_precision_recall_curve(
...    preds, target, num_labels=3, thresholds=None
... )
>>> precision  
[tensor([0.5000, 0.5000, 1.0000, 1.0000]), tensor([0.6667, 0.5000, 0.0000, 1.0000]),
 tensor([0.7500, 1.0000, 1.0000, 1.0000])]
>>> recall  
[tensor([1.0000, 0.5000, 0.5000, 0.0000]), tensor([1.0000, 0.5000, 0.0000, 0.0000]),
 tensor([1.0000, 0.6667, 0.3333, 0.0000])]
>>> thresholds  
[tensor([0.0500, 0.4500, 0.7500]), tensor([0.5500, 0.6500, 0.7500]),
 tensor([0.0500, 0.3500, 0.7500])]
>>> multilabel_precision_recall_curve(
...     preds, target, num_labels=3, thresholds=5
... )  
(tensor([[0.5000, 0.5000, 1.0000, 1.0000, 0.0000, 1.0000],
         [0.5000, 0.6667, 0.6667, 0.0000, 0.0000, 1.0000],
         [0.7500, 1.0000, 1.0000, 1.0000, 0.0000, 1.0000]]),
 tensor([[1.0000, 0.5000, 0.5000, 0.5000, 0.0000, 0.0000],
         [1.0000, 1.0000, 1.0000, 0.0000, 0.0000, 0.0000],
         [1.0000, 0.6667, 0.3333, 0.3333, 0.0000, 0.0000]]),
 tensor([0.0000, 0.2500, 0.5000, 0.7500, 1.0000]))