/
curves.py
153 lines (129 loc) · 5.56 KB
/
curves.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
# Copyright 2022 Google LLC
#
# 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.
# ==============================================================================
"""An interpreters for generating data for ROC and PR curves."""
from collections.abc import Sequence
from typing import cast, Optional
from lit_nlp.api import components as lit_components
from lit_nlp.api import dataset as lit_dataset
from lit_nlp.api import model as lit_model
from lit_nlp.api import types
from lit_nlp.lib import utils as lit_utils
import numpy as np
from sklearn import metrics
JsonDict = types.JsonDict
IndexedInput = types.IndexedInput
Spec = types.Spec
# The config key for specifying model output to use for calculations.
TARGET_PREDICTION_KEY = 'Prediction field'
# The config key for specifying the class label to use for calculations.
TARGET_LABEL_KEY = 'Label'
# They field name in the interpreter output that contains ROC curve data.
ROC_DATA = 'roc_data'
# They field name in the interpreter output that contains PR curve data.
PR_DATA = 'pr_data'
class CurvesInterpreter(lit_components.Interpreter):
"""Returns data for rendering ROC and Precision-Recall curves."""
def run(self,
inputs: Sequence[JsonDict],
model: lit_model.Model,
dataset: lit_dataset.Dataset,
model_outputs: Optional[Sequence[JsonDict]] = None,
config: Optional[JsonDict] = None):
if not config:
raise ValueError('Curves required config parameters but received none.')
if (target_label := config.get(TARGET_LABEL_KEY)) is None:
raise ValueError(
f'The config \'{TARGET_LABEL_KEY}\' field should contain the positive'
f' class label.')
# Find the prediction field key in the model output to use for calculations.
output_spec = model.output_spec()
if TARGET_PREDICTION_KEY in config:
predictions_key: str = config[TARGET_PREDICTION_KEY]
elif len(pred_keys := self._find_supported_pred_keys(output_spec)) == 1:
predictions_key: str = pred_keys[0]
else:
raise ValueError(
'Unable to determine prediction field. Please provide one via the'
f' "{TARGET_PREDICTION_KEY}" field in the CallConfig or update the'
' model spec to output a single MulticlassPreds field.'
)
if not inputs:
return {ROC_DATA: [], PR_DATA: []}
# Run prediction if needed:
if model_outputs is None:
model_outputs = list(model.predict(inputs))
# Get scores for the target label.
pred_spec = output_spec.get(predictions_key)
if not isinstance(pred_spec, types.MulticlassPreds):
raise TypeError(
f'Expected {predictions_key} to be a MulticlassPreds field, but got a'
f' {type(pred_spec).__name__}'
)
labels = pred_spec.vocab
target_index = labels.index(target_label)
scores = [o[predictions_key][target_index] for o in model_outputs]
# Get ground truth for the target label.
parent_key = pred_spec.parent
ground_truth_list = []
for ex in inputs:
ground_truth_label = ex[parent_key]
ground_truth = 1.0 if ground_truth_label == target_label else 0.0
ground_truth_list.append(ground_truth)
# Compute ROC curve data.
x, y, _ = metrics.roc_curve(ground_truth_list, scores)
roc_data = list(zip(np.nan_to_num(x), np.nan_to_num(y)))
roc_data.sort(key=lambda x: x[0])
# Compute PR curve data.
x, y, _ = metrics.precision_recall_curve(ground_truth_list, scores)
pr_data = list(zip(np.nan_to_num(x), np.nan_to_num(y)))
pr_data.sort(key=lambda x: x[0])
# Create and return the result.
return {ROC_DATA: roc_data, PR_DATA: pr_data}
def is_compatible(
self, model: lit_model.Model, dataset: lit_dataset.Dataset
) -> bool:
"""True if using a classification model and dataset has ground truth."""
output_spec = model.output_spec()
supported_keys = self._find_supported_pred_keys(output_spec)
has_parents = all(
cast(types.MulticlassPreds, output_spec[key]).parent in dataset.spec()
for key in supported_keys
)
return bool(supported_keys) and has_parents
def config_spec(self) -> types.Spec:
# If a model is a multiclass classifier, a user can specify which
# class label to use for plotting the curves. If the label is not
# specified then the label with index 0 is used by default.
return {
TARGET_LABEL_KEY: types.CategoryLabel(),
TARGET_PREDICTION_KEY: types.SingleFieldMatcher(
spec='output', types=['MulticlassPreds'], required=False
),
}
def meta_spec(self) -> types.Spec:
return {ROC_DATA: types.CurveDataPoints(), PR_DATA: types.CurveDataPoints()}
def _find_supported_pred_keys(self, output_spec: types.Spec) -> list[str]:
"""Returns the list of supported prediction keys in the model output.
Args:
output_spec: The model output specification.
Returns:
The list of keys.
"""
all_keys = lit_utils.find_spec_keys(output_spec, types.MulticlassPreds)
supported_keys = [
k for k in all_keys
if cast(types.MulticlassPreds, output_spec[k]).parent
]
return supported_keys