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 unittest
from unittest.mock import patch
import numpy as np
from tests.utils import TestTestBase
from baycomp.single import CorrelatedTTest, Posterior, two_on_single
class PosteriorTest(unittest.TestCase):
def test_sample(self):
posterior = Posterior(42, 0, 1, nsamples=10)
np.testing.assert_almost_equal(posterior.sample, np.full((10, ), 42))
posterior = Posterior(42, 2, 3, nsamples=10)
def mockstandardt(df, nsamples):
return np.arange(nsamples) * df
with patch("numpy.random.standard_t", mockstandardt):
np.testing.assert_almost_equal(
posterior.sample,
42 + np.sqrt(2) * np.arange(10) * 3)
def test_probs_var_0(self):
# var=0, df=1, rope=2
np.testing.assert_equal(Posterior(42, 0, 1, 2).probs(), [0, 0, 1])
np.testing.assert_equal(Posterior(2, 0, 1, 2).probs(), [0, 1, 0])
np.testing.assert_equal(Posterior(.5, 0, 1, 2).probs(), [0, 1, 0])
np.testing.assert_equal(Posterior(0, 0, 1, 2).probs(), [0, 1, 0])
np.testing.assert_equal(Posterior(-.5, 0, 1, 2).probs(), [0, 1, 0])
np.testing.assert_equal(Posterior(-2, 0, 1, 2).probs(), [0, 1, 0])
np.testing.assert_equal(Posterior(-42, 0, 1, 2).probs(), [1, 0, 0])
# var=0, df=1, rope=0
np.testing.assert_equal(Posterior(42, 0, 1).probs(), [0, 1])
np.testing.assert_equal(Posterior(0, 0, 1).probs(), [0.5, 0.5])
np.testing.assert_equal(Posterior(-42, 0, 1).probs(), [1, 0])
def test_probs(self):
def mockcdf(x, a_df, a_loc, a_scale):
self.assertEqual(a_df, 1)
self.assertEqual(a_loc, 3)
self.assertAlmostEqual(a_scale, 5)
# 0.25 below rope (or 0)
# 0.4 below + between
return 0.25 if x <= 0 else 0.4
with patch("scipy.stats.t.cdf", mockcdf):
posterior = Posterior(mean=3, var=25, df=1, rope=0)
np.testing.assert_almost_equal(
posterior.probs(), [0.25, 0.75])
posterior = Posterior(mean=3, var=25, df=1, rope=2)
np.testing.assert_almost_equal(
posterior.probs(), [0.25, 0.15, 0.6])
class CorrelatedTTestTest(TestTestBase):
def test_new(self):
x = np.array([4, 2, 6])
y = np.array([5, 7, 3])
with patch.object(CorrelatedTTest,
"compute_statistics", return_value=(1, 2, 3)) as cs:
posterior = CorrelatedTTest(x, y, rope=6, runs=3, names=("a", "b"))
cs.assert_called_with(x, y, 3)
self.assertEqual(posterior.mean, 1)
self.assertEqual(posterior.var, 2)
self.assertEqual(posterior.df, 3)
self.assertEqual(posterior.rope, 6)
self.assertEqual(posterior.meanx, 4)
self.assertEqual(posterior.meany, 5)
self.assertEqual(posterior.names, ("a", "b"))
def test_new_checks_errors(self):
x = np.array([4, 2, 6])
y = np.array([5, 7, 3])
with patch("baycomp.single.check_args") as ca:
CorrelatedTTest(x, y, 12)
ca.assert_called_with(x, y, 12)
self.assertRaises(ValueError, CorrelatedTTest, x, y, runs=0)
self.assertRaises(ValueError, CorrelatedTTest, x, y, runs=2)
self.assertRaises(ValueError, CorrelatedTTest, x, y, runs=-1)
self.assertRaises(ValueError, CorrelatedTTest, x, y, runs=0.5)
def test_compute_statistics(self):
x = np.array([4, 2, 6])
y = np.array([5, 7, 3])
self.assertEqual(
CorrelatedTTest.compute_statistics(x, y, 1),
(1, (4 ** 2 + 4 ** 2) / 2 * (1 / 3 + 1 / 2), 2))
x = np.array([4, 2, 6, 4])
y = np.array([5, 7, 3, 5])
self.assertEqual(
CorrelatedTTest.compute_statistics(x, y, 2),
(1, (4 ** 2 + 4 ** 2) / 3 * (1 / 4 + 1), 3))
x = np.ones(10)
y = np.zeros(10)
self.assertEqual(
CorrelatedTTest.compute_statistics(x, y),
(-1, 0, 9))
def test_sample(self):
x = np.array([42, 42, 42])
y = np.zeros(3)
np.testing.assert_almost_equal(
CorrelatedTTest.sample(x, y, 1, nsamples=10),
np.full((10, ), -42))
def mockstandardt(df, nsamples):
return np.arange(nsamples) * df
with patch("numpy.random.standard_t", mockstandardt),\
patch.object(CorrelatedTTest, "compute_statistics",
return_value=(1, 2, 3)):
np.testing.assert_almost_equal(
CorrelatedTTest.sample(x, y, 1, nsamples=10),
1 + np.sqrt(2) * np.arange(10) * 3)
def test_probs(self):
x, y = object(), object()
self.assert_forwards(CorrelatedTTest, "probs", x, y, 2, 1)
def test_plot(self):
x, y = object(), object()
names = object()
self.assert_forwards(
CorrelatedTTest, "plot", x, y, 2, 1, names=names,
new_args=(x, y, 2, 1), new_kwargs={}, meth_args=(names,))
class TwoOnSingleTest(unittest.TestCase):
def test_two_on_single(self):
x, y = object(), object()
names = ("a, b")
with patch("baycomp.single.call_shortcut") as mockshortcut:
two_on_single(x, y, 0.5, 10, plot=True, names=names)
mockshortcut.assert_called_with(CorrelatedTTest, x, y, 0.5,
plot=True, names=names, runs=10)
if __name__ == "__main__":
unittest.main()
|
