Skip to content

[skip ci] Add doctest for Precision, Recall and FBeta metrics #2340

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
sdesrozis merged 6 commits into from
Dec 7, 2021
Merged

[skip ci] Add doctest for Precision, Recall and FBeta metrics #2340

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
ade561b
Add doctest for Precision, Recall and FBeta metrics
Dec 3, 2021
48613e0
Fix lint for fbeta
sdesrozis Dec 5, 2021
20e2efe
Update for lint
sdesrozis Dec 5, 2021
51ee29f
Update for lint
sdesrozis Dec 5, 2021
7c20c9a
Merge branch 'master' into doctest_fbeta
sdesrozis Dec 6, 2021
74f7fa7
Merge branch 'master' into doctest_fbeta
sdesrozis Dec 6, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
94 changes: 94 additions & 0 deletions ignite/metrics/fbeta.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,10 @@ def Fbeta(

where :math:`\beta` is a positive real factor.

- ``update`` must receive output of the form ``(y_pred, y)`` or ``{'y_pred': y_pred, 'y': y}``.
- `y_pred` must be in the following shape (batch_size, num_categories, ...) or (batch_size, ...).
- `y` must be in the following shape (batch_size, ...).

Args:
beta: weight of precision in harmonic mean
average: if True, F-beta score is computed as the unweighted average (across all classes
Expand All @@ -40,6 +44,96 @@ def Fbeta(

Returns:
MetricsLambda, F-beta metric

Examples:

Binary case

.. testcode:: 1

P = Precision(average=False)
R = Recall(average=False)
metric = Fbeta(beta=1.0, precision=P, recall=R)
metric.attach(default_evaluator, "f-beta")
y_true = torch.Tensor([1, 0, 1, 1, 0, 1])
y_pred = torch.Tensor([1, 0, 1, 0, 1, 1])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["f-beta"])

.. testoutput:: 1

0.7499...

Multiclass case

.. testcode:: 2

P = Precision(average=False)
R = Recall(average=False)
metric = Fbeta(beta=1.0, precision=P, recall=R)
metric.attach(default_evaluator, "f-beta")
y_true = torch.Tensor([2, 0, 2, 1, 0, 1]).long()
y_pred = torch.Tensor([
[0.0266, 0.1719, 0.3055],
[0.6886, 0.3978, 0.8176],
[0.9230, 0.0197, 0.8395],
[0.1785, 0.2670, 0.6084],
[0.8448, 0.7177, 0.7288],
[0.7748, 0.9542, 0.8573],
])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["f-beta"])

.. testoutput:: 2

0.5222...

F-beta can be computed for each class as done below:

.. testcode:: 3

P = Precision(average=False)
R = Recall(average=False)
metric = Fbeta(beta=1.0, average=False, precision=P, recall=R)
metric.attach(default_evaluator, "f-beta")
y_true = torch.Tensor([2, 0, 2, 1, 0, 1]).long()
y_pred = torch.Tensor([
[0.0266, 0.1719, 0.3055],
[0.6886, 0.3978, 0.8176],
[0.9230, 0.0197, 0.8395],
[0.1785, 0.2670, 0.6084],
[0.8448, 0.7177, 0.7288],
[0.7748, 0.9542, 0.8573],
])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["f-beta"])

.. testoutput:: 3

tensor([0.5000, 0.6667, 0.4000], dtype=torch.float64)

The elements of `y` and `y_pred` should have 0 or 1 values. Thresholding of predictions can
be done as below:

.. testcode:: 4

def thresholded_output_transform(output):
y_pred, y = output
y_pred = torch.round(y_pred)
return y_pred, y

P = Precision(average=False, output_transform=thresholded_output_transform)
R = Recall(average=False, output_transform=thresholded_output_transform)
metric = Fbeta(beta=1.0, precision=P, recall=R)
metric.attach(default_evaluator, "f-beta")
y_true = torch.Tensor([1, 0, 1, 1, 0, 1])
y_pred = torch.Tensor([0.6, 0.2, 0.9, 0.4, 0.7, 0.65])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["f-beta"])

.. testoutput:: 4

0.7499...
"""
if not (beta > 0):
raise ValueError(f"Beta should be a positive integer, but given {beta}")
Expand Down
100 changes: 98 additions & 2 deletions ignite/metrics/precision.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,17 +87,113 @@ class Precision(_BasePrecisionRecall):
default, CPU.

Examples:

Binary case

.. testcode:: 1

metric = Precision(average=False)
metric.attach(default_evaluator, "precision")
y_true = torch.Tensor([1, 0, 1, 1, 0, 1])
y_pred = torch.Tensor([1, 0, 1, 0, 1, 1])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["precision"])

.. testoutput:: 1

0.75

Multiclass case

.. testcode:: 2

metric = Precision(average=False)
metric.attach(default_evaluator, "precision")
y_true = torch.Tensor([2, 0, 2, 1, 0, 1]).long()
y_pred = torch.Tensor([
[0.0266, 0.1719, 0.3055],
[0.6886, 0.3978, 0.8176],
[0.9230, 0.0197, 0.8395],
[0.1785, 0.2670, 0.6084],
[0.8448, 0.7177, 0.7288],
[0.7748, 0.9542, 0.8573],
])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["precision"])

.. testoutput:: 2

tensor([0.5000, 1.0000, 0.3333], dtype=torch.float64)

Precision can be computed as the unweighted average across all classes:

.. testcode:: 3

metric = Precision(average=True)
metric.attach(default_evaluator, "precision")
y_true = torch.Tensor([2, 0, 2, 1, 0, 1]).long()
y_pred = torch.Tensor([
[0.0266, 0.1719, 0.3055],
[0.6886, 0.3978, 0.8176],
[0.9230, 0.0197, 0.8395],
[0.1785, 0.2670, 0.6084],
[0.8448, 0.7177, 0.7288],
[0.7748, 0.9542, 0.8573],
])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["precision"])

.. testoutput:: 3

0.6111...

Multilabel case

.. testcode:: 4

metric = Precision(average=True, is_multilabel=True)
metric.attach(default_evaluator, "precision")
y_true = torch.Tensor([
[0, 0, 1, 0, 1],
[0, 0, 0, 0, 1],
[0, 0, 0, 0, 1],
[1, 0, 0, 0, 1],
[0, 1, 1, 0, 1],
])
y_pred = torch.Tensor([
[1, 1, 0, 0, 0],
[1, 0, 1, 0, 0],
[1, 0, 0, 0, 0],
[1, 0, 1, 1, 1],
[1, 1, 0, 0, 1],
])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["precision"])

.. testoutput:: 4

0.23333...

In binary and multilabel cases, the elements of `y` and `y_pred` should have 0 or 1 values. Thresholding of
predictions can be done as below:

.. code-block:: python
.. testcode:: 5

def thresholded_output_transform(output):
y_pred, y = output
y_pred = torch.round(y_pred)
return y_pred, y

precision = Precision(output_transform=thresholded_output_transform)
metric = Precision(average=False, output_transform=thresholded_output_transform)
metric.attach(default_evaluator, "precision")
y_true = torch.Tensor([1, 0, 1, 1, 0, 1])
y_pred = torch.Tensor([0.6, 0.2, 0.9, 0.4, 0.7, 0.65])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["precision"])

.. testoutput:: 5

0.75

In multilabel cases, average parameter should be True. However, if user would like to compute F1 metric, for
example, average parameter should be False. This can be done as shown below:
Expand Down
100 changes: 98 additions & 2 deletions ignite/metrics/recall.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,17 +34,113 @@ class Recall(_BasePrecisionRecall):
default, CPU.

Examples:

Binary case

.. testcode:: 1

metric = Recall(average=False)
metric.attach(default_evaluator, "recall")
y_true = torch.Tensor([1, 0, 1, 1, 0, 1])
y_pred = torch.Tensor([1, 0, 1, 0, 1, 1])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["recall"])

.. testoutput:: 1

0.75

Multiclass case

.. testcode:: 2

metric = Recall(average=False)
metric.attach(default_evaluator, "recall")
y_true = torch.Tensor([2, 0, 2, 1, 0, 1]).long()
y_pred = torch.Tensor([
[0.0266, 0.1719, 0.3055],
[0.6886, 0.3978, 0.8176],
[0.9230, 0.0197, 0.8395],
[0.1785, 0.2670, 0.6084],
[0.8448, 0.7177, 0.7288],
[0.7748, 0.9542, 0.8573],
])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["recall"])

.. testoutput:: 2

tensor([0.5000, 0.5000, 0.5000], dtype=torch.float64)

Precision can be computed as the unweighted average across all classes:

.. testcode:: 3

metric = Recall(average=True)
metric.attach(default_evaluator, "recall")
y_true = torch.Tensor([2, 0, 2, 1, 0, 1]).long()
y_pred = torch.Tensor([
[0.0266, 0.1719, 0.3055],
[0.6886, 0.3978, 0.8176],
[0.9230, 0.0197, 0.8395],
[0.1785, 0.2670, 0.6084],
[0.8448, 0.7177, 0.7288],
[0.7748, 0.9542, 0.8573],
])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["recall"])

.. testoutput:: 3

0.5

Multilabel case

.. testcode:: 4

metric = Recall(average=True, is_multilabel=True)
metric.attach(default_evaluator, "recall")
y_true = torch.Tensor([
[0, 0, 1, 0, 1],
[0, 0, 0, 0, 1],
[0, 0, 0, 0, 1],
[1, 0, 0, 0, 1],
[0, 1, 1, 0, 1],
])
y_pred = torch.Tensor([
[1, 1, 0, 0, 0],
[1, 0, 1, 0, 0],
[1, 0, 0, 0, 0],
[1, 0, 1, 1, 1],
[1, 1, 0, 0, 1],
])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["recall"])

.. testoutput:: 4

0.33333...

In binary and multilabel cases, the elements of `y` and `y_pred` should have 0 or 1 values. Thresholding of
predictions can be done as below:

.. code-block:: python
.. testcode:: 5

def thresholded_output_transform(output):
y_pred, y = output
y_pred = torch.round(y_pred)
return y_pred, y

recall = Recall(output_transform=thresholded_output_transform)
metric = Recall(average=False, output_transform=thresholded_output_transform)
metric.attach(default_evaluator, "recall")
y_true = torch.Tensor([1, 0, 1, 1, 0, 1])
y_pred = torch.Tensor([0.6, 0.2, 0.9, 0.4, 0.7, 0.65])
state = default_evaluator.run([[y_pred, y_true]])
print(state.metrics["recall"])

.. testoutput:: 5

0.75

In multilabel cases, average parameter should be True. However, if user would like to compute F1 metric, for
example, average parameter should be False. This can be done as shown below:
Expand Down