-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathtest_reliability.py
145 lines (117 loc) · 3.54 KB
/
test_reliability.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
import numpy as np
import pandas as pd
from ChildProject.metrics import (
gamma,
segments_to_annotation,
segments_to_grid,
grid_to_vector,
vectors_to_annotation_task,
conf_matrix,
)
def test_gamma():
segments = pd.read_csv("tests/data/gamma.csv")
value = gamma(segments, "speaker_type", alpha=3, beta=1, precision_level=0.01)
assert 0.39 <= value <= 0.44
def test_segments_to_grid():
segments = pd.read_csv("tests/data/grid.csv")
grid_both = segments_to_grid(
segments, 0, 10, 1, "speaker_type", ["CHI", "FEM"], overlap=True, none=True
)
grid_none_only = segments_to_grid(
segments, 0, 10, 1, "speaker_type", ["CHI", "FEM"], overlap=False, none=True
)
grid_overlap_only = segments_to_grid(
segments, 0, 10, 1, "speaker_type", ["CHI", "FEM"], overlap=True, none=False
)
grid_bare = segments_to_grid(
segments, 0, 10, 1, "speaker_type", ["CHI", "FEM"], overlap=False, none=False
)
truth = np.array(
[
[1, 0, 0, 0],
[1, 0, 0, 0],
[0, 1, 0, 0],
[0, 1, 0, 0],
[0, 0, 0, 1],
[0, 0, 0, 1],
[1, 1, 1, 0],
[1, 1, 1, 0],
[0, 1, 0, 0],
[0, 0, 0, 1],
]
)
np.testing.assert_array_equal(grid_both, truth)
np.testing.assert_array_equal(grid_none_only, np.delete(truth, -2, 1))
np.testing.assert_array_equal(grid_overlap_only, truth[:, :-1])
np.testing.assert_array_equal(grid_bare, truth[:, :-2])
def test_grid_to_vectors():
segments = pd.read_csv("tests/data/grid.csv")
grid = segments_to_grid(
segments, 0, 10, 1, "speaker_type", ["CHI", "FEM"], overlap=True, none=True
)
vector = grid_to_vector(grid, ["CHI", "FEM", "overlap", "none"])
truth = np.array(
[
"CHI",
"CHI",
"FEM",
"FEM",
"none",
"none",
"overlap",
"overlap",
"FEM",
"none",
]
)
np.testing.assert_array_equal(vector, truth)
def test_conf_matrix():
segments = pd.read_csv("tests/data/confmatrix.csv")
categories = ["CHI", "FEM"]
confmat = conf_matrix(
segments_to_grid(
segments[segments["set"] == "Bob"],
0,
20,
1,
"speaker_type",
categories,
overlap=True,
none=True,
),
segments_to_grid(
segments[segments["set"] == "Alice"],
0,
20,
1,
"speaker_type",
categories,
overlap=True,
none=True,
),
)
truth = np.array([[5, 5, 0, 0], [0, 2, 0, 3], [0, 0, 0, 0], [0, 0, 0, 5]])
np.testing.assert_array_equal(confmat, truth)
def test_alpha():
segments = pd.read_csv("tests/data/alpha.csv")
categories = list(segments["speaker_type"].unique())
sets = list(segments["set"].unique())
vectors = [
grid_to_vector(
segments_to_grid(
segments[segments["set"] == s],
0,
segments["segment_offset"].max(),
1,
"speaker_type",
categories,
overlap=True,
none=True,
),
categories + ["overlap", "none"],
)
for s in sets
]
task = vectors_to_annotation_task(*vectors, drop=["none"])
alpha = task.alpha()
assert np.isclose(alpha, 0.743421052632, rtol=0.001, atol=0.0001)