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
|
# -*- coding: utf-8 -*-
"""MultiLabelCM module."""
from __future__ import division
from typing import List, Set, Any, Union, Optional
from .errors import pycmVectorError, pycmMultiLabelError
from .params import *
from .cm import ConfusionMatrix
import numpy
class MultiLabelCM():
"""
Multilabel confusion matrix class.
>>> mlcm = MultiLabelCM([{'dog'}, {'cat', 'dog'}], [{'cat'}, {'cat'}])
>>> mlcm.actual_vector_multihot
[[0, 1], [1, 1]]
>>> mlcm.predict_vector_multihot
[[1, 0], [1, 0]]
"""
def __init__(
self,
actual_vector: Union[List[Set[Any]], numpy.ndarray],
predict_vector: Union[List[Set[Any]], numpy.ndarray],
sample_weight: Optional[Union[List[float], numpy.ndarray]] = None,
classes: Optional[List[Any]] = None) -> None:
"""
Init method.
:param actual_vector: actual vector
:param predict_vector: vector of predictions
:param sample_weight: sample weights list
:param classes: ordered labels of classes
"""
self.actual_vector = actual_vector
self.actual_vector_multihot = []
self.predict_vector = predict_vector
self.predict_vector_multihot = []
self.weights = None
self.classes = None
self.classwise_cms = {}
self.samplewise_cms = {}
__mlcm_vector_handler__(
self,
actual_vector,
predict_vector,
sample_weight,
classes)
__mlcm_assign_classes__(self, classes)
__mlcm_vectors_filter__(self)
def get_cm_by_class(self, class_name: Any) -> "ConfusionMatrix":
"""
Return confusion matrices based on classes.
:param class_name: target class name for confusion matrix
"""
if class_name not in self.classwise_cms:
try:
class_index = self.classes.index(class_name)
except ValueError:
raise pycmMultiLabelError(INVALID_CLASS_NAME_ERROR)
actual_vector_temp = []
predict_vector_temp = []
for item1, item2 in zip(
self.actual_vector_multihot, self.predict_vector_multihot):
actual_vector_temp.append(item1[class_index])
predict_vector_temp.append(item2[class_index])
cm = ConfusionMatrix(
actual_vector_temp,
predict_vector_temp,
sample_weight=self.weights)
self.classwise_cms[class_name] = cm
return cm
return self.classwise_cms[class_name]
def get_cm_by_sample(self, index: int) -> "ConfusionMatrix":
"""
Return confusion matrices based on samples.
:param index: sample index for confusion matrix
"""
if index < 0 or index >= len(self.actual_vector):
raise pycmMultiLabelError(VECTOR_INDEX_ERROR)
if index not in self.samplewise_cms:
cm = ConfusionMatrix(
self.actual_vector_multihot[index],
self.predict_vector_multihot[index],
sample_weight=self.weights)
self.samplewise_cms[index] = cm
return cm
return self.samplewise_cms[index]
def __str__(self) -> str:
"""Multilabel confusion matrix object string representation method."""
return self.__repr__()
def __repr__(self) -> str:
"""Multilabel confusion matrix object representation method."""
return "pycm.MultiLabelCM(classes: " + str(self.classes) + ")"
def __len__(self) -> int:
"""Multilabel confusion matrix object length method."""
return len(self.classes)
def __mlcm_vector_handler__(
mlcm: "MultiLabelCM",
actual_vector: Union[List[Set[Any]], numpy.ndarray],
predict_vector: Union[List[Set[Any]], numpy.ndarray],
sample_weight: Union[List[float], numpy.ndarray],
classes: List[Any]) -> None:
"""
Handle multilabel object conditions for vectors.
:param mlcm: multilabel confusion matrix
:param actual_vector: actual vector
:param predict_vector: vector of predictions
:param sample_weight: sample weights list
:param classes: ordered labels of classes
"""
if not isinstance(actual_vector, (list, numpy.ndarray)) or not \
isinstance(predict_vector, (list, numpy.ndarray)):
raise pycmVectorError(VECTOR_TYPE_ERROR)
if len(actual_vector) != len(predict_vector):
raise pycmVectorError(VECTOR_SIZE_ERROR)
if len(actual_vector) == 0 or len(predict_vector) == 0:
raise pycmVectorError(VECTOR_EMPTY_ERROR)
if not all(map(lambda x: isinstance(x, set), actual_vector)):
raise pycmVectorError(NOT_ALL_SET_VECTOR_ERROR)
if not all(map(lambda x: isinstance(x, set), predict_vector)):
raise pycmVectorError(NOT_ALL_SET_VECTOR_ERROR)
if classes is not None and len(set(classes)) != len(classes):
raise pycmVectorError(VECTOR_UNIQUE_CLASS_ERROR)
if isinstance(sample_weight, (list, numpy.ndarray)):
mlcm.weights = sample_weight
def __mlcm_assign_classes__(
mlcm: "MultiLabelCM",
classes: List[Any]) -> None:
"""
Assign multilabel object class.
:param mlcm: multilabel confusion matrix
:param classes: ordered labels of classes
"""
mlcm.classes = classes
if classes is None:
mlcm.classes = sorted(
list(
set.union(
*mlcm.actual_vector,
*mlcm.predict_vector)))
def __mlcm_vectors_filter__(mlcm: "MultiLabelCM") -> None:
"""
Normalize multilabel object vectors.
:param mlcm: multilabel confusion matrix
"""
mlcm.actual_vector_multihot = [__set_to_multihot__(
x, mlcm.classes) for x in mlcm.actual_vector]
mlcm.predict_vector_multihot = [__set_to_multihot__(
x, mlcm.classes) for x in mlcm.predict_vector]
def __set_to_multihot__(input_set: Set[Any], classes: List[Any]) -> List[int]:
"""
Convert a set into a multi-hot vector based in classes.
:param input_set: input set
:param classes: ordered labels of classes
"""
result = [0] * len(classes)
for i, x in enumerate(classes):
if x in input_set:
result[i] = 1
return result
|
