-
Notifications
You must be signed in to change notification settings - Fork 584
/
test_bayesian_network_structure_learning.py
356 lines (253 loc) · 10.9 KB
/
test_bayesian_network_structure_learning.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
355
356
# test_bayesian_network_structure_learning.py
# Contact: Jacob Schreiber <jmschreiber91@gmail.com>
import numpy
import torch
import pytest
from pomegranate.bayesian_network import BayesianNetwork
from pomegranate.bayesian_network import _from_structure
from pomegranate.bayesian_network import _learn_structure
from pomegranate.bayesian_network import _categorical_exact
from pomegranate.bayesian_network import _categorical_chow_liu
from pomegranate.distributions import Categorical
from pomegranate.distributions import ConditionalCategorical
from nose.tools import assert_raises
from numpy.testing import assert_array_almost_equal
def assert_tuple_equal(x, y):
assert len(x) == len(y)
for i in range(len(x)):
assert len(x[i]) == len(y[i])
if len(x[i]) > 0:
for j in range(len(x[i])):
assert x[i][j] == y[i][j]
@pytest.fixture
def X():
return [[1, 2, 0, 0],
[0, 0, 1, 0],
[1, 1, 1, 0],
[1, 2, 1, 1],
[1, 1, 0, 1],
[0, 1, 0, 1],
[0, 1, 0, 0],
[1, 0, 1, 1],
[1, 1, 0, 1],
[0, 2, 1, 0],
[0, 0, 0, 1]]
@pytest.fixture
def X_masked(X):
mask = torch.tensor(numpy.array([
[False, True, True, True ],
[True, True, False, False],
[False, False, False, True ],
[True, True, True, True ],
[False, True, False, False],
[True, True, True, False],
[False, False, False, True ],
[True, False, True, True ],
[True, True, True, False],
[True, True, True, True ],
[True, False, True, False]]))
X = torch.tensor(numpy.array(X))
return torch.masked.MaskedTensor(X, mask=mask)
@pytest.fixture
def w():
return torch.tensor([1.0, 0.3, 1.0, 0.0, 0.0, 2.1, 3.1, 1.0, 1.2, 1.1, 2.0])
###
def test_categorical_chow_liu(X):
structure = _categorical_chow_liu(X)
assert_tuple_equal(structure, ((), (0,), (0,), (0,)))
structure = _categorical_chow_liu(X, root=1)
assert_tuple_equal(structure, ((1,), (), (1,), (1,)))
structure = _categorical_chow_liu(X, root=2)
assert_tuple_equal(structure, ((2,), (2,), (), (2,)))
def test_categorical_chow_liu_weighted(X, w):
structure = _categorical_chow_liu(X, w)
assert_tuple_equal(structure, ((), (0,), (0,), (0,)))
structure = _categorical_chow_liu(X, w, root=1)
assert_tuple_equal(structure, ((1,), (), (1,), (1,)))
structure = _categorical_chow_liu(X, w, root=2)
assert_tuple_equal(structure, ((2,), (2,), (), (2,)))
def test_categorical_chow_liu_large():
numpy.random.seed(0)
X = numpy.random.randint(3, size=(50, 10))
structure = _categorical_chow_liu(X)
assert_tuple_equal(structure, ((), (0,), (1,), (7,), (0,), (0,), (7,), (2,),
(7,), (0,)))
structure = _categorical_chow_liu(X, root=1)
assert_tuple_equal(structure, ((1,), (), (1,), (7,), (0,), (0,), (7,), (2,),
(7,), (0,)))
structure = _categorical_chow_liu(X, root=2)
assert_tuple_equal(structure, ((1,), (2,), (), (7,), (0,), (0,), (7,), (2,),
(7,), (0,)))
def test_categorical_chow_liu_large_pseudocount():
numpy.random.seed(0)
X = numpy.random.randint(3, size=(50, 10))
structure = _categorical_chow_liu(X, pseudocount=10)
assert_tuple_equal(structure, ((), (2,), (4,), (2,), (0,), (3,), (3,), (2,),
(2,), (6,)))
structure = _categorical_chow_liu(X, root=1, pseudocount=10)
assert_tuple_equal(structure, ((4,), (), (1,), (2,), (2,), (3,), (3,), (2,),
(2,), (6,)))
structure = _categorical_chow_liu(X, root=2, pseudocount=10)
assert_tuple_equal(structure, ((4,), (2,), (), (2,), (2,), (3,), (3,), (2,),
(2,), (6,)))
def test_categorical_chow_liu_raises(X, w):
assert_raises(ValueError, _categorical_chow_liu, X, w, -1)
assert_raises(TypeError, _categorical_chow_liu, X, w, [1.2, 1.1])
assert_raises(ValueError, _categorical_chow_liu, X, w, None, -1)
assert_raises(ValueError, _categorical_chow_liu, X, w, None, 0.3)
assert_raises(ValueError, _categorical_chow_liu, X, w, None, [1, 2])
w = torch.tensor(w)
assert_raises(ValueError, _categorical_chow_liu, X, w.unsqueeze(0))
assert_raises(ValueError, _categorical_chow_liu, X, w.unsqueeze(1))
X = numpy.array(X)
n, d = X.shape
assert_raises(ValueError, _categorical_chow_liu, X + 0.3, w)
assert_raises(ValueError, _categorical_chow_liu, X.reshape(1, n, d), w)
assert_raises(ValueError, _categorical_chow_liu, X - 1, w)
###
def test_categorical_exact(X):
structure = _categorical_exact(X)
assert_tuple_equal(structure, ((), (0,), (0, 1), ()))
structure = _categorical_exact(X, max_parents=1)
assert_tuple_equal(structure, ((), (0,), (), ()))
def test_categorical_exact_weighted(X, w):
structure = _categorical_exact(X, w)
assert_tuple_equal(structure, ((), (0,), (0, 1), ()))
structure = _categorical_exact(X, w, max_parents=1)
assert_tuple_equal(structure, ((), (0,), (), ()))
def test_categorical_exact_exclude_parents(X):
exclude_parents = ((), (2,), (), (1,))
structure = _categorical_exact(X, exclude_parents=exclude_parents)
assert_tuple_equal(structure, ((), (), (0,), (0, 2)))
structure = _categorical_exact(X, exclude_parents=exclude_parents,
max_parents=1)
assert_tuple_equal(structure, ((), (0,), (), ()))
exclude_parents = ((), (2,), (), (0, 1))
structure = _categorical_exact(X, exclude_parents=exclude_parents)
assert_tuple_equal(structure, ((2, 3), (), (), (2,)))
def test_categorical_exact_large():
numpy.random.seed(0)
X = numpy.random.randint(3, size=(50, 10))
structure = _categorical_exact(X)
assert_tuple_equal(structure, ((1, 2), (), (), (), (0, 1, 2), (), (),
(), (0, 1, 2, 4), ()))
structure = _categorical_exact(X, max_parents=1)
assert_tuple_equal(structure, ((), (), (), (), (), (), (), (2,), (), ()))
structure = _categorical_exact(X, max_parents=2)
assert_tuple_equal(structure, ((), (), (), (), (), (), (), (0, 1), (),
()))
structure = _categorical_exact(X, max_parents=3)
assert_tuple_equal(structure, ((1, 2), (), (), (), (0, 1, 2), (), (), (),
(), ()))
structure = _categorical_exact(X, max_parents=4)
assert_tuple_equal(structure, ((1, 2), (), (), (), (0, 1, 2), (), (), (),
(0, 1, 2, 4), ()))
def test_categorical_exact_large_pseudocount():
numpy.random.seed(0)
X = numpy.random.randint(3, size=(50, 8))
structure = _categorical_exact(X, pseudocount=10)
assert_tuple_equal(structure, ((), (), (), (), (), (), (), ()))
structure = _categorical_exact(X, pseudocount=5)
assert_tuple_equal(structure, ((), (), (), (), (), (), (), ()))
structure = _categorical_exact(X, pseudocount=2)
assert_tuple_equal(structure, ((), (), (), (), (), (), (), ()))
structure = _categorical_exact(X, pseudocount=1)
assert_tuple_equal(structure, ((), (), (), (), (), (), (), ()))
structure = _categorical_exact(X, pseudocount=0.1)
assert_tuple_equal(structure, ((), (), (), (), (), (), (), ()))
structure = _categorical_exact(X, pseudocount=1e-8)
assert_tuple_equal(structure, ((), (), (), (), (), (0,), (), ()))
###
def test_categorical_learn_structure_chow_liu():
numpy.random.seed(0)
X = numpy.random.randint(3, size=(50, 7))
w = numpy.abs(numpy.random.randn(50))
structure1 = _categorical_chow_liu(X)
structure2 = _learn_structure(X, algorithm='chow-liu')
assert_tuple_equal(structure1, structure2)
structure1 = _categorical_chow_liu(X, root=1)
structure2 = _learn_structure(X, root=1, algorithm='chow-liu')
assert_tuple_equal(structure1, structure2)
structure1 = _categorical_chow_liu(X, pseudocount=50)
structure2 = _learn_structure(X, pseudocount=50, algorithm='chow-liu')
assert_tuple_equal(structure1, structure2)
structure1 = _categorical_chow_liu(X, w)
structure2 = _learn_structure(X, w, algorithm='chow-liu')
assert_tuple_equal(structure1, structure2)
def test_categorical_learn_structure_exact():
numpy.random.seed(0)
X = numpy.random.randint(3, size=(50, 7))
w = numpy.abs(numpy.random.randn(50))
structure1 = _categorical_exact(X)
structure2 = _learn_structure(X, algorithm='exact')
assert_tuple_equal(structure1, structure2)
structure1 = _categorical_exact(X, max_parents=1)
structure2 = _learn_structure(X, max_parents=1, algorithm='exact')
assert_tuple_equal(structure1, structure2)
structure1 = _categorical_exact(X, pseudocount=50)
structure2 = _learn_structure(X, pseudocount=50, algorithm='exact')
assert_tuple_equal(structure1, structure2)
structure1 = _categorical_exact(X, w)
structure2 = _learn_structure(X, w, algorithm='exact')
assert_tuple_equal(structure1, structure2)
###
def test_categorical_from_structure(X):
structure = ((), (0,), (1, 3), ())
distributions = _from_structure(X, structure=structure)
model = BayesianNetwork(distributions, structure=structure)
assert isinstance(model.distributions[0], Categorical)
assert isinstance(model.distributions[1], ConditionalCategorical)
assert isinstance(model.distributions[2], ConditionalCategorical)
assert isinstance(model.distributions[3], Categorical)
p0 = model.distributions[0].probs
p1 = model.distributions[1].probs[0]
p2 = model.distributions[2].probs[0]
p3 = model.distributions[3].probs
assert_array_almost_equal(p0, [[0.454545, 0.545455]])
assert_array_almost_equal(p1, [[0.4 , 0.4 , 0.2 ],
[0.166667, 0.5 , 0.333333]])
assert_array_almost_equal(p2, [[[0.0000, 1.0000],
[0.5000, 0.5000]],
[[0.5000, 0.5000],
[1.0000, 0.0000]],
[[0.5000, 0.5000],
[0.0000, 1.0000]]])
assert_array_almost_equal(p3, [[0.4545, 0.5455]], 4)
def test_categorical_from_structure_weighted(X, w):
structure = ((), (0,), (1, 3), ())
distributions = _from_structure(X, sample_weight=w, structure=structure)
model = BayesianNetwork(distributions, structure=structure)
assert isinstance(model.distributions[0], Categorical)
assert isinstance(model.distributions[1], ConditionalCategorical)
assert isinstance(model.distributions[2], ConditionalCategorical)
assert isinstance(model.distributions[3], Categorical)
p0 = model.distributions[0].probs
p1 = model.distributions[1].probs[0]
p2 = model.distributions[2].probs[0]
p3 = model.distributions[3].probs
assert_array_almost_equal(p0, [[0.6719, 0.3281]], 4)
assert_array_almost_equal(p1, [[0.2674, 0.6047, 0.1279],
[0.2381, 0.5238, 0.2381]], 4)
assert_array_almost_equal(p2, [[[0.0000, 1.0000],
[0.6667, 0.3333]],
[[0.7561, 0.2439],
[1.0000, 0.0000]],
[[0.4762, 0.5238],
[0.5000, 0.5000]]], 4)
assert_array_almost_equal(p3, [[0.5078, 0.4922]], 4)
def test_categorical_from_structure_null(X):
structure = ((), (), (), ())
distributions = _from_structure(X, structure=structure)
model = BayesianNetwork(distributions, structure=structure)
assert isinstance(model.distributions[0], Categorical)
assert isinstance(model.distributions[1], Categorical)
assert isinstance(model.distributions[2], Categorical)
assert isinstance(model.distributions[3], Categorical)
p0 = model.distributions[0].probs
p1 = model.distributions[1].probs
p2 = model.distributions[2].probs
p3 = model.distributions[3].probs
assert_array_almost_equal(p0, [[0.454545, 0.545455]])
assert_array_almost_equal(p1, [[0.272727, 0.454545, 0.272727]])
assert_array_almost_equal(p2, [[0.5455, 0.4545]], 4)
assert_array_almost_equal(p3, [[0.4545, 0.5455]], 4)