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
|
# Authors: The MNE-Python contributors.
# License: BSD-3-Clause
# Copyright the MNE-Python contributors.
from functools import partial
from itertools import product
import numpy as np
import pytest
import scipy.stats
from numpy.testing import assert_allclose, assert_array_almost_equal, assert_array_less
import mne
from mne.stats.parametric import _map_effects, f_mway_rm, f_threshold_mway_rm
# hardcoded external test results, manually transferred
test_external = {
# SPSS, manually conducted analysis
"spss_fvals": np.array([2.568, 0.240, 1.756]),
"spss_pvals_uncorrected": np.array([0.126, 0.788, 0.186]),
"spss_pvals_corrected": np.array([0.126, 0.784, 0.192]),
# R 15.2
# data generated using this code http://goo.gl/7UcKb
"r_fvals": np.array([2.567619, 0.24006, 1.756380]),
"r_pvals_uncorrected": np.array([0.12557, 0.78776, 0.1864]),
# and https://gist.github.com/dengemann/5539403
"r_fvals_3way": np.array(
[
0.74783999999999995, # A
0.20895, # B
0.21378, # A:B
0.99404000000000003, # C
0.094039999999999999, # A:C
0.11685, # B:C
2.78749,
]
), # A:B:C
"r_fvals_1way": np.array([0.67571999999999999]),
}
def generate_data(n_subjects, n_conditions):
"""Generate testing data."""
rng = np.random.RandomState(42)
data = rng.randn(n_subjects * n_conditions).reshape(n_subjects, n_conditions)
return data
def test_map_effects():
"""Test ANOVA effects parsing."""
selection, names = _map_effects(n_factors=2, effects="A")
assert names == ["A"]
selection, names = _map_effects(n_factors=2, effects=["A", "A:B"])
assert names == ["A", "A:B"]
selection, names = _map_effects(n_factors=3, effects="A*B")
assert names == ["A", "B", "A:B"]
# XXX this might be wrong?
selection, names = _map_effects(n_factors=3, effects="A*C")
assert names == ["A", "B", "A:B", "C", "A:C"]
pytest.raises(ValueError, _map_effects, n_factors=2, effects="C")
pytest.raises(ValueError, _map_effects, n_factors=27, effects="all")
def test_f_twoway_rm():
"""Test 2-way anova."""
rng = np.random.RandomState(42)
iter_params = product([4, 10], [2, 15], [4, 6, 8], ["A", "B", "A:B"], [False, True])
_effects = {4: [2, 2], 6: [2, 3], 8: [2, 4]}
for params in iter_params:
n_subj, n_obs, n_levels, effects, correction = params
data = rng.random_sample([n_subj, n_levels, n_obs])
fvals, pvals = f_mway_rm(
data, _effects[n_levels], effects, correction=correction
)
assert (fvals >= 0).all()
if pvals.any():
assert ((0 <= pvals) & (1 >= pvals)).all()
n_effects = len(_map_effects(n_subj, effects)[0])
assert fvals.size == n_obs * n_effects
if n_effects == 1: # test for principle of least surprise ...
assert fvals.ndim == 1
fvals_ = f_threshold_mway_rm(n_subj, _effects[n_levels], effects)
assert (fvals_ >= 0).all()
assert fvals_.size == n_effects
# check time-frequency input
n_subj, n_freqs, n_times, n_levels = (5, 10, 101, 4)
data = rng.random_sample([n_subj, n_levels, n_freqs, n_times])
fvals, pvals = f_mway_rm(data, _effects[n_levels])
assert fvals.shape[1:] == pvals.shape[1:] == (n_freqs, n_times)
data = rng.random_sample([n_subj, n_levels, 1])
pytest.raises(
ValueError,
f_mway_rm,
data,
_effects[n_levels],
effects="C",
correction=correction,
)
data = rng.random_sample([n_subj, n_levels, n_obs, 3])
# check for dimension handling
f_mway_rm(data, _effects[n_levels], effects, correction=correction)
# now check against external software results
test_data = generate_data(n_subjects=20, n_conditions=6)
fvals, pvals = f_mway_rm(test_data, [2, 3])
assert_array_almost_equal(fvals, test_external["spss_fvals"], 3)
assert_array_almost_equal(pvals, test_external["spss_pvals_uncorrected"], 3)
assert_array_almost_equal(fvals, test_external["r_fvals"], 4)
assert_array_almost_equal(pvals, test_external["r_pvals_uncorrected"], 3)
_, pvals = f_mway_rm(test_data, [2, 3], correction=True)
assert_array_almost_equal(pvals, test_external["spss_pvals_corrected"], 3)
test_data = generate_data(n_subjects=20, n_conditions=8)
fvals, _ = f_mway_rm(test_data, [2, 2, 2])
assert_array_almost_equal(fvals, test_external["r_fvals_3way"], 5)
fvals, _ = f_mway_rm(test_data, [8], "A")
assert_array_almost_equal(fvals, test_external["r_fvals_1way"], 5)
@pytest.mark.parametrize(
"kind, kwargs",
[
("1samp", {}),
("ind", {}), # equal_var=True is the default
("ind", dict(equal_var=True)),
("ind", dict(equal_var=False)),
],
)
@pytest.mark.parametrize(
"sigma",
(
0.0,
1e-3,
),
)
@pytest.mark.parametrize("seed", [0, 42, 1337])
def test_ttest_equiv(kind, kwargs, sigma, seed):
"""Test t-test equivalence."""
rng = np.random.RandomState(seed)
def theirs(*a, **kw):
f = getattr(scipy.stats, f"ttest_{kind}")
if kind == "1samp":
func = partial(f, popmean=0, **kwargs)
else:
func = partial(f, **kwargs)
return func(*a, **kw)[0]
ours = partial(getattr(mne.stats, f"ttest_{kind}_no_p"), sigma=sigma, **kwargs)
X = rng.randn(3, 4, 5)
if kind == "ind":
X = [X, rng.randn(30, 4, 5)] # should differ based on equal_var
got = ours(*X)
want = theirs(*X)
else:
got = ours(X)
want = theirs(X)
if sigma == 0.0:
assert_allclose(got, want, rtol=1e-7, atol=1e-6)
else:
assert not np.allclose(got, want, rtol=1e-7, atol=1e-6)
# should mostly be similar, but uniformly smaller because we add
# something to the divisor (var)
assert_allclose(got, want, rtol=2e-1, atol=1e-2)
assert_array_less(np.abs(got), np.abs(want))
|
