-
Notifications
You must be signed in to change notification settings - Fork 415
/
test_dissim.py
153 lines (122 loc) · 6.55 KB
/
test_dissim.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
"""
Tests for dissimilarity measures
"""
import unittest
import numpy as np
from sklearn.utils._testing import assert_array_equal
from kmodes.util.dissim import matching_dissim, euclidean_dissim, ng_dissim
from kmodes.util.dissim import jaccard_dissim_binary, jaccard_dissim_label
# pylint: disable=no-self-use,pointless-statement
class TestDissimilarityMeasures(unittest.TestCase):
def test_matching_dissim(self):
a = np.array([[0, 1, 2, 0, 1, 2]])
b = np.array([[0, 1, 2, 0, 1, 0]])
self.assertEqual(1, matching_dissim(a, b))
a = np.array([[np.NaN, 1, 2, 0, 1, 2]])
b = np.array([[0, 1, 2, 0, 1, 0]])
self.assertEqual(2, matching_dissim(a, b))
a = np.array([['a', 'b', 'c', 'd']])
b = np.array([['a', 'b', 'c', 'd'], ['d', 'c', 'b', 'a']])
assert_array_equal(np.array([0, 4]), matching_dissim(a, b))
def test_jaccard_dissim_binary(self):
a = np.array([[0, 1, 1, 0, 1, 1]])
b = np.array([[0, 1, 1, 0, 1, 0]])
self.assertEqual(0.25, jaccard_dissim_binary(a, b))
a = np.array([[0, 1, 1, 0, 1, 1]])
b = np.array([[0, np.NaN, 1, 0, 1, 0]])
with self.assertRaises(ValueError):
jaccard_dissim_binary(a, b)
# test where values are non binary but also not having np.NaN
a = np.array([[0, 1, 2, 0, 1, 2]])
b = np.array([[0, 1, 2, 0, 1, 0]])
with self.assertRaises(ValueError):
jaccard_dissim_binary(a, b)
# test for dissimilarity = 0: sets are the same
a = np.array([[1, 1, 0, 1, 1, 0]])
b = np.array([[1, 1, 0, 1, 1, 0]])
self.assertEqual(0, jaccard_dissim_binary(a, b))
# test for dissimilarity = 1: sets are completely different
a = np.array([[0, 0, 1, 0, 0, 1]])
b = np.array([[1, 1, 0, 1, 1, 0]])
self.assertEqual(1, jaccard_dissim_binary(a, b))
def test_jaccard_dissim_label(self):
a = np.array([[0, 1, 2, 0, 1, 2]])
b = np.array([[0, 1, 2, 0, 3, 0]])
self.assertEqual(0.25, jaccard_dissim_label(a, b))
a = np.array([[np.NaN, 1, 2, 0, 1, 2]])
b = np.array([[0, 1, 2, 0, 1, 0]])
with self.assertRaises(ValueError):
jaccard_dissim_label(a, b)
# test for dissimilarity = 0: sets are the same
a = np.array([[1, 2, 0, 3, 1, 0]])
b = np.array([[1, 2, 0, 3, 1, 0]])
self.assertEqual(0, jaccard_dissim_label(a, b))
# test for dissimilarity = 1: sets are completely different
a = np.array([[1, 2, 0, 3, 1, 0]])
b = np.array([[5, 4, 6, 7, 8, 9]])
self.assertEqual(1, jaccard_dissim_label(a, b))
def test_euclidian_dissim(self):
a = np.array([[0., 1., 2., 0., 1., 2.]])
b = np.array([[3., 1., 3., 0., 1., 0.]])
self.assertEqual(14., euclidean_dissim(a, b))
a = np.array([[np.NaN, 1., 2., 0., 1., 2.]])
b = np.array([[3., 1., 3., 0., 1., 0.]])
with self.assertRaises(ValueError):
euclidean_dissim(a, b)
def test_ng_dissim(self):
X = np.array([[0, 1, 2, 0, 1, 2], [0, 1, 2, 0, 1, 1]])
centroids = X
membship = np.array([[1, 0], [0, 1]])
assert_array_equal(np.array([0., 1.]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([1., 0.]),
ng_dissim(centroids, X[1], X=X, membship=membship))
# Unit test for initialization (i.e., same as matching_dissim)
membship = np.array([[0, 0], [0, 0]])
mdiss_00 = matching_dissim(np.array([X[0]]), np.array([X[0]]))[0]
mdiss_01 = matching_dissim(np.array([X[0]]), np.array([X[1]]))[0]
mdiss_11 = matching_dissim(np.array([X[1]]), np.array([X[1]]))[0]
assert_array_equal(np.array([mdiss_00, mdiss_01]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([mdiss_01, mdiss_11]),
ng_dissim(centroids, X[1], X=X, membship=membship))
# Unit test for NaN
X = np.array([[np.NaN, 1, 2, 0, 1, 2], [0, 1, 2, 0, 1, 1]])
centroids = X
membship = np.array([[1, 0], [0, 1]])
assert_array_equal(np.array([1., 2.]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([2., 0.]),
ng_dissim(centroids, X[1], X=X, membship=membship))
# Unit test for initialization with NaN(i.e., same as matching_dissim)
membship = np.array([[0, 0], [0, 0]])
mdiss_00 = matching_dissim(np.array([X[0]]), np.array([X[0]]))[0]
mdiss_01 = matching_dissim(np.array([X[0]]), np.array([X[1]]))[0]
mdiss_11 = matching_dissim(np.array([X[1]]), np.array([X[1]]))[0]
assert_array_equal(np.array([mdiss_00, mdiss_01]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([mdiss_01, mdiss_11]),
ng_dissim(centroids, X[1], X=X, membship=membship))
X = np.array([['a', 'b', 'c', 'd'], ['a', 'b', 'e', 'd'], ['d', 'c', 'b', 'a']])
centroids = np.array([['a', 'b', 'c', 'd'], ['d', 'c', 'b', 'a']])
membship = np.array([[1, 1, 0], [0, 0, 1]])
assert_array_equal(np.array([0.5, 4.]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([1., 4.]),
ng_dissim(centroids, X[1], X=X, membship=membship))
assert_array_equal(np.array([4., 0.]),
ng_dissim(centroids, X[2], X=X, membship=membship))
# Unit test for initialization (i.e., same as matching_dissim)
membship = np.array([[0, 0, 0], [0, 0, 0]])
mdiss_00 = matching_dissim(np.array([X[0]]), np.array([X[0]]))[0]
mdiss_01 = matching_dissim(np.array([X[0]]), np.array([X[1]]))[0]
mdiss_11 = matching_dissim(np.array([X[1]]), np.array([X[1]]))[0]
mdiss_02 = matching_dissim(np.array([X[0]]), np.array([X[2]]))[0]
mdiss_12 = matching_dissim(np.array([X[0]]), np.array([X[2]]))[0]
mdiss_22 = matching_dissim(np.array([X[2]]), np.array([X[2]]))[0]
assert_array_equal(np.array([mdiss_00, mdiss_02]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([mdiss_01, mdiss_12]),
ng_dissim(centroids, X[1], X=X, membship=membship))
assert_array_equal(np.array([mdiss_12, mdiss_22]),
ng_dissim(centroids, X[2], X=X, membship=membship))