/
f_beta.py
354 lines (295 loc) · 16.9 KB
/
f_beta.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
# Copyright The PyTorch Lightning team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import Optional
import torch
from torch import Tensor
from torchmetrics.functional.classification.stat_scores import _reduce_stat_scores, _stat_scores_update
from torchmetrics.utilities.enums import AverageMethod as AvgMethod
from torchmetrics.utilities.enums import MDMCAverageMethod
def _safe_divide(num: Tensor, denom: Tensor) -> Tensor:
"""prevent zero division."""
denom[denom == 0.0] = 1
return num / denom
def _fbeta_compute(
tp: Tensor,
fp: Tensor,
tn: Tensor,
fn: Tensor,
beta: float,
ignore_index: Optional[int],
average: str,
mdmc_average: Optional[str],
) -> Tensor:
"""Computes f_beta metric from stat scores: true positives, false positives, true negatives, false negatives.
Args:
tp: True positives
fp: False positives
tn: True negatives
fn: False negatives
beta: The parameter `beta` (which determines the weight of recall in the combined score)
ignore_index: Integer specifying a target class to ignore. If given, this class index does not contribute
to the returned score, regardless of reduction method
average: Defines the reduction that is applied
mdmc_average: Defines how averaging is done for multi-dimensional multi-class inputs (on top of the
``average`` parameter)
Example:
>>> from torchmetrics.functional.classification.stat_scores import _stat_scores_update
>>> target = torch.tensor([0, 1, 2, 0, 1, 2])
>>> preds = torch.tensor([0, 2, 1, 0, 0, 1])
>>> tp, fp, tn, fn = _stat_scores_update(
... preds,
... target,
... reduce='micro',
... num_classes=3,
... )
>>> _fbeta_compute(tp, fp, tn, fn, beta=0.5, ignore_index=None, average='micro', mdmc_average=None)
tensor(0.3333)
"""
if average == AvgMethod.MICRO and mdmc_average != MDMCAverageMethod.SAMPLEWISE:
mask = tp >= 0
precision = _safe_divide(tp[mask].sum().float(), (tp[mask] + fp[mask]).sum())
recall = _safe_divide(tp[mask].sum().float(), (tp[mask] + fn[mask]).sum())
else:
precision = _safe_divide(tp.float(), tp + fp)
recall = _safe_divide(tp.float(), tp + fn)
num = (1 + beta**2) * precision * recall
denom = beta**2 * precision + recall
denom[denom == 0.0] = 1.0 # avoid division by 0
# if classes matter and a given class is not present in both the preds and the target,
# computing the score for this class is meaningless, thus they should be ignored
if average == AvgMethod.NONE and mdmc_average != MDMCAverageMethod.SAMPLEWISE:
# a class is not present if there exists no TPs, no FPs, and no FNs
meaningless_indeces = torch.nonzero((tp | fn | fp) == 0).cpu()
if ignore_index is None:
ignore_index = meaningless_indeces
else:
ignore_index = torch.unique(torch.cat((meaningless_indeces, torch.tensor([[ignore_index]]))))
if ignore_index is not None:
if average not in (AvgMethod.MICRO, AvgMethod.SAMPLES) and mdmc_average == MDMCAverageMethod.SAMPLEWISE:
num[..., ignore_index] = -1
denom[..., ignore_index] = -1
elif average not in (AvgMethod.MICRO, AvgMethod.SAMPLES):
num[ignore_index, ...] = -1
denom[ignore_index, ...] = -1
if average == AvgMethod.MACRO and mdmc_average != MDMCAverageMethod.SAMPLEWISE:
cond = (tp + fp + fn == 0) | (tp + fp + fn == -3)
num = num[~cond]
denom = denom[~cond]
return _reduce_stat_scores(
numerator=num,
denominator=denom,
weights=None if average != AvgMethod.WEIGHTED else tp + fn,
average=average,
mdmc_average=mdmc_average,
)
def fbeta_score(
preds: Tensor,
target: Tensor,
beta: float = 1.0,
average: str = "micro",
mdmc_average: Optional[str] = None,
ignore_index: Optional[int] = None,
num_classes: Optional[int] = None,
threshold: float = 0.5,
top_k: Optional[int] = None,
multiclass: Optional[bool] = None,
) -> Tensor:
r"""
Computes f_beta metric.
.. math::
F_{\beta} = (1 + \beta^2) * \frac{\text{precision} * \text{recall}}
{(\beta^2 * \text{precision}) + \text{recall}}
Works with binary, multiclass, and multilabel data.
Accepts probabilities or logits from a model output or integer class values in prediction.
Works with multi-dimensional preds and target.
If preds and target are the same shape and preds is a float tensor, we use the ``self.threshold`` argument
to convert into integer labels. This is the case for binary and multi-label logits or probabilities.
If preds has an extra dimension as in the case of multi-class scores we perform an argmax on ``dim=1``.
The reduction method (how the precision scores are aggregated) is controlled by the
``average`` parameter, and additionally by the ``mdmc_average`` parameter in the
multi-dimensional multi-class case. Accepts all inputs listed in :ref:`pages/classification:input types`.
Args:
preds: Predictions from model (probabilities, logits or labels)
target: Ground truth values
beta: beta coefficient
average:
Defines the reduction that is applied. Should be one of the following:
- ``'micro'`` [default]: Calculate the metric globally, across all samples and classes.
- ``'macro'``: Calculate the metric for each class separately, and average the
metrics across classes (with equal weights for each class).
- ``'weighted'``: Calculate the metric for each class separately, and average the
metrics across classes, weighting each class by its support (``tp + fn``).
- ``'none'`` or ``None``: Calculate the metric for each class separately, and return
the metric for every class.
- ``'samples'``: Calculate the metric for each sample, and average the metrics
across samples (with equal weights for each sample).
.. note:: What is considered a sample in the multi-dimensional multi-class case
depends on the value of ``mdmc_average``.
.. note:: If ``'none'`` and a given class doesn't occur in the ``preds`` or ``target``,
the value for the class will be ``nan``.
mdmc_average:
Defines how averaging is done for multi-dimensional multi-class inputs (on top of the
``average`` parameter). Should be one of the following:
- ``None`` [default]: Should be left unchanged if your data is not multi-dimensional
multi-class.
- ``'samplewise'``: In this case, the statistics are computed separately for each
sample on the ``N`` axis, and then averaged over samples.
The computation for each sample is done by treating the flattened extra axes ``...``
(see :ref:`pages/classification:input types`) as the ``N`` dimension within the sample,
and computing the metric for the sample based on that.
- ``'global'``: In this case the ``N`` and ``...`` dimensions of the inputs
(see :ref:`pages/classification:input types`)
are flattened into a new ``N_X`` sample axis, i.e. the inputs are treated as if they
were ``(N_X, C)``. From here on the ``average`` parameter applies as usual.
ignore_index:
Integer specifying a target class to ignore. If given, this class index does not contribute
to the returned score, regardless of reduction method. If an index is ignored, and ``average=None``
or ``'none'``, the score for the ignored class will be returned as ``nan``.
num_classes:
Number of classes. Necessary for ``'macro'``, ``'weighted'`` and ``None`` average methods.
threshold:
Threshold for transforming probability or logit predictions to binary (0,1) predictions, in the case
of binary or multi-label inputs. Default value of 0.5 corresponds to input being probabilities.
top_k:
Number of highest probability or logit score predictions considered to find the correct label,
relevant only for (multi-dimensional) multi-class inputs. The
default value (``None``) will be interpreted as 1 for these inputs.
Should be left at default (``None``) for all other types of inputs.
multiclass:
Used only in certain special cases, where you want to treat inputs as a different type
than what they appear to be. See the parameter's
:ref:`documentation section <pages/classification:using the multiclass parameter>`
for a more detailed explanation and examples.
Return:
The shape of the returned tensor depends on the ``average`` parameter
- If ``average in ['micro', 'macro', 'weighted', 'samples']``, a one-element tensor will be returned
- If ``average in ['none', None]``, the shape will be ``(C,)``, where ``C`` stands for the number
of classes
Example:
>>> from torchmetrics.functional import fbeta_score
>>> target = torch.tensor([0, 1, 2, 0, 1, 2])
>>> preds = torch.tensor([0, 2, 1, 0, 0, 1])
>>> fbeta_score(preds, target, num_classes=3, beta=0.5)
tensor(0.3333)
"""
allowed_average = list(AvgMethod)
if average not in allowed_average:
raise ValueError(f"The `average` has to be one of {allowed_average}, got {average}.")
if mdmc_average is not None and MDMCAverageMethod.from_str(mdmc_average) is None:
raise ValueError(f"The `mdmc_average` has to be one of {list(MDMCAverageMethod)}, got {mdmc_average}.")
if average in [AvgMethod.MACRO, AvgMethod.WEIGHTED, AvgMethod.NONE] and (not num_classes or num_classes < 1):
raise ValueError(f"When you set `average` as {average}, you have to provide the number of classes.")
if num_classes and ignore_index is not None and (not 0 <= ignore_index < num_classes or num_classes == 1):
raise ValueError(f"The `ignore_index` {ignore_index} is not valid for inputs with {num_classes} classes")
reduce = AvgMethod.MACRO if average in [AvgMethod.WEIGHTED, AvgMethod.NONE] else average
tp, fp, tn, fn = _stat_scores_update(
preds,
target,
reduce=reduce,
mdmc_reduce=mdmc_average,
threshold=threshold,
num_classes=num_classes,
top_k=top_k,
multiclass=multiclass,
ignore_index=ignore_index,
)
return _fbeta_compute(tp, fp, tn, fn, beta, ignore_index, average, mdmc_average)
def f1_score(
preds: Tensor,
target: Tensor,
beta: float = 1.0,
average: str = "micro",
mdmc_average: Optional[str] = None,
ignore_index: Optional[int] = None,
num_classes: Optional[int] = None,
threshold: float = 0.5,
top_k: Optional[int] = None,
multiclass: Optional[bool] = None,
) -> Tensor:
"""Computes F1 metric. F1 metrics correspond to a equally weighted average of the precision and recall scores.
Works with binary, multiclass, and multilabel data.
Accepts probabilities or logits from a model output or integer class values in prediction.
Works with multi-dimensional preds and target.
If preds and target are the same shape and preds is a float tensor, we use the ``self.threshold`` argument
to convert into integer labels. This is the case for binary and multi-label probabilities or logits.
If preds has an extra dimension as in the case of multi-class scores we perform an argmax on ``dim=1``.
The reduction method (how the precision scores are aggregated) is controlled by the
``average`` parameter, and additionally by the ``mdmc_average`` parameter in the
multi-dimensional multi-class case. Accepts all inputs listed in :ref:`pages/classification:input types`.
Args:
preds: Predictions from model (probabilities, logits or labels)
target: Ground truth values
beta: it is ignored
average:
Defines the reduction that is applied. Should be one of the following:
- ``'micro'`` [default]: Calculate the metric globally, across all samples and classes.
- ``'macro'``: Calculate the metric for each class separately, and average the
metrics across classes (with equal weights for each class).
- ``'weighted'``: Calculate the metric for each class separately, and average the
metrics across classes, weighting each class by its support (``tp + fn``).
- ``'none'`` or ``None``: Calculate the metric for each class separately, and return
the metric for every class.
- ``'samples'``: Calculate the metric for each sample, and average the metrics
across samples (with equal weights for each sample).
.. note:: What is considered a sample in the multi-dimensional multi-class case
depends on the value of ``mdmc_average``.
.. note:: If ``'none'`` and a given class doesn't occur in the ``preds`` or ``target``,
the value for the class will be ``nan``.
mdmc_average:
Defines how averaging is done for multi-dimensional multi-class inputs (on top of the
``average`` parameter). Should be one of the following:
- ``None`` [default]: Should be left unchanged if your data is not multi-dimensional multi-class.
- ``'samplewise'``: In this case, the statistics are computed separately for each
sample on the ``N`` axis, and then averaged over samples.
The computation for each sample is done by treating the flattened extra axes ``...``
(see :ref:`pages/classification:input types`) as the ``N`` dimension within the sample,
and computing the metric for the sample based on that.
- ``'global'``: In this case the ``N`` and ``...`` dimensions of the inputs
(see :ref:`pages/classification:input types`)
are flattened into a new ``N_X`` sample axis, i.e. the inputs are treated as if they
were ``(N_X, C)``. From here on the ``average`` parameter applies as usual.
ignore_index:
Integer specifying a target class to ignore. If given, this class index does not contribute
to the returned score, regardless of reduction method. If an index is ignored, and ``average=None``
or ``'none'``, the score for the ignored class will be returned as ``nan``.
num_classes:
Number of classes. Necessary for ``'macro'``, ``'weighted'`` and ``None`` average methods.
threshold:
Threshold for transforming probability or logit predictions to binary (0,1) predictions, in the case
of binary or multi-label inputs. Default value of 0.5 corresponds to input being probabilities.
top_k:
Number of highest probability or logit score predictions considered to find the correct label,
relevant only for (multi-dimensional) multi-class inputs. The
default value (``None``) will be interpreted as 1 for these inputs.
Should be left at default (``None``) for all other types of inputs.
multiclass:
Used only in certain special cases, where you want to treat inputs as a different type
than what they appear to be. See the parameter's
:ref:`documentation section <pages/classification:using the multiclass parameter>`
for a more detailed explanation and examples.
Return:
The shape of the returned tensor depends on the ``average`` parameter
- If ``average in ['micro', 'macro', 'weighted', 'samples']``, a one-element tensor will be returned
- If ``average in ['none', None]``, the shape will be ``(C,)``, where ``C`` stands for the number
of classes
Example:
>>> from torchmetrics.functional import f1_score
>>> target = torch.tensor([0, 1, 2, 0, 1, 2])
>>> preds = torch.tensor([0, 2, 1, 0, 0, 1])
>>> f1_score(preds, target, num_classes=3)
tensor(0.3333)
"""
return fbeta_score(
preds, target, 1.0, average, mdmc_average, ignore_index, num_classes, threshold, top_k, multiclass
)