def test_sem(self):
# This is not in R, so used: sqrt(var(testcase)*3/4) / sqrt(3)
y = mstats.sem(self.testcase)
assert_almost_equal(y, 0.6454972244)
n = self.testcase.count()
assert_allclose(mstats.sem(self.testcase, ddof=0) * np.sqrt(n/(n-2)),
mstats.sem(self.testcase, ddof=2))
def test_nd_input(self):
x = np.array((-2,-1,0,1,2,3)*4)**2
x_2d = np.vstack([x] * 2).T
for func in [mstats.normaltest, mstats.skewtest, mstats.kurtosistest]:
res_1d = func(x)
res_2d = func(x_2d)
assert_allclose(res_2d[0], [res_1d[0]] * 2)
assert_allclose(res_2d[1], [res_1d[1]] * 2)
def test_skewtest(self):
# this test is for 1D data
for n in self.get_n():
if n > 8:
x, y, xm, ym = self.generate_xy_sample(n)
r = stats.skewtest(x)
rm = stats.mstats.skewtest(xm)
assert_allclose(r[0], rm[0], rtol=1e-15)
def test_sem(self):
# This is not in R, so used: sqrt(var(testcase)*3/4) / sqrt(3)
# Note, differs from stats.sem return due to different ddof (backwards
# compat reasons).
y = mstats.sem(self.testcase)
assert_almost_equal(y, 0.55901699437494745)
n = self.testcase.count()
assert_allclose(mstats.sem(self.testcase, ddof=0) * np.sqrt(n/(n-2)),
mstats.sem(self.testcase, ddof=2))
def test_gmean(self):
for n in self.get_n():
x, y, xm, ym = self.generate_xy_sample(n)
r = stats.gmean(abs(x))
rm = stats.mstats.gmean(abs(xm))
assert_allclose(r, rm, rtol=1e-13)
r = stats.gmean(abs(y))
rm = stats.mstats.gmean(abs(ym))
assert_allclose(r, rm, rtol=1e-13)
def test_normaltest(self):
np.seterr(over='raise')
for n in self.get_n():
if n > 8:
with warnings.catch_warnings():
warnings.filterwarnings('ignore', category=UserWarning)
x, y, xm, ym = self.generate_xy_sample(n)
r = stats.normaltest(x)
rm = stats.mstats.normaltest(xm)
assert_allclose(np.asarray(r), np.asarray(rm))
def test_sem(self):
# example from stats.sem doc
a = np.arange(20).reshape(5,4)
am = np.ma.array(a)
r = stats.sem(a,ddof=1)
rm = stats.mstats.sem(am, ddof=1)
assert_allclose(r, 2.82842712, atol=1e-5)
assert_allclose(rm, 2.82842712, atol=1e-5)
for n in self.get_n():
x, y, xm, ym = self.generate_xy_sample(n)
assert_almost_equal(stats.mstats.sem(xm, axis=None, ddof=0),
stats.sem(x, axis=None, ddof=0), decimal=13)
assert_almost_equal(stats.mstats.sem(ym, axis=None, ddof=0),
stats.sem(y, axis=None, ddof=0), decimal=13)
assert_almost_equal(stats.mstats.sem(xm, axis=None, ddof=1),
stats.sem(x, axis=None, ddof=1), decimal=13)
assert_almost_equal(stats.mstats.sem(ym, axis=None, ddof=1),
stats.sem(y, axis=None, ddof=1), decimal=13)
def test_maskedarray_input(self):
# Add some masked values, test result doesn't change
x = np.array((-2, -1, 0, 1, 2, 3) * 4) ** 2
xm = np.ma.array(np.r_[np.inf, x, 10], mask=np.r_[True, [False] * x.size, True])
assert_allclose(mstats.normaltest(xm), stats.normaltest(x))
assert_allclose(mstats.skewtest(xm), stats.skewtest(x))
assert_allclose(mstats.kurtosistest(xm), stats.kurtosistest(x))
def test_axis_None(self):
# Test axis=None (equal to axis=0 for 1-D input)
x = np.array((-2,-1,0,1,2,3)*4)**2
assert_allclose(mstats.normaltest(x, axis=None), mstats.normaltest(x))
assert_allclose(mstats.skewtest(x, axis=None), mstats.skewtest(x))
assert_allclose(mstats.kurtosistest(x, axis=None),
mstats.kurtosistest(x))
def test_maskedarray_input(self):
# Add some masked values, test result doesn't change
x = np.array((-2,-1,0,1,2,3)*4)**2
xm = np.ma.array(np.r_[np.inf, x, 10],
mask=np.r_[True, [False] * x.size, True])
assert_allclose(mstats.normaltest(xm), stats.normaltest(x))
assert_allclose(mstats.skewtest(xm), stats.skewtest(x))
assert_allclose(mstats.kurtosistest(xm), stats.kurtosistest(x))
def test_hilbert_theoretical_long_signals(self):
"""
Tests the output of the hilbert function with regard to amplitude and
phase of long test signals
"""
# Performing test using all pad types
for padding in ['nextpow', 'none', 16384]:
h = elephant.signal_processing.hilbert(
self.long_signals, N=padding)
phase = np.angle(h.magnitude)
amplitude = np.abs(h.magnitude)
real_value = np.real(h.magnitude)
# The real part should be equal to the original long_signals
assert_array_almost_equal(
real_value,
self.long_signals.magnitude,
decimal=14)
# Test only in the middle half of the array (border effects)
ind1 = int(len(h.times) / 4)
ind2 = int(3 * len(h.times) / 4)
# Calculate difference in phase between signal and original phase
# and use smaller of any two phase differences
phasediff = np.abs(phase[ind1:ind2, :] - self.phase[ind1:ind2, :])
phasediff[phasediff >= np.pi] = \
2 * np.pi - phasediff[phasediff >= np.pi]
# Calculate difference in amplitude between signal and original
# amplitude
amplitudediff = \
amplitude[ind1:ind2, :] - self.amplitude[ind1:ind2, :]
#
assert_allclose(phasediff, 0, atol=0.1)
assert_allclose(amplitudediff, 0, atol=0.5)
def test_hilbert_theoretical_long_signals(self):
"""
Tests the output of the hilbert function with regard to amplitude and
phase of long test signals
"""
# Performing test using all pad types
for padding in ['nextpow', 'none', 16384]:
h = elephant.signal_processing.hilbert(
self.long_signals, N=padding)
phase = np.angle(h.magnitude)
amplitude = np.abs(h.magnitude)
real_value = np.real(h.magnitude)
# The real part should be equal to the original long_signals
assert_array_almost_equal(
real_value,
self.long_signals.magnitude,
decimal=14)
# Test only in the middle half of the array (border effects)
ind1 = int(len(h.times) / 4)
ind2 = int(3 * len(h.times) / 4)
# Calculate difference in phase between signal and original phase
# and use smaller of any two phase differences
phasediff = np.abs(phase[ind1:ind2, :] - self.phase[ind1:ind2, :])
phasediff[phasediff >= np.pi] = \
2 * np.pi - phasediff[phasediff >= np.pi]
# Calculate difference in amplitude between signal and original
# amplitude
amplitudediff = \
amplitude[ind1:ind2, :] - self.amplitude[ind1:ind2, :]
#
assert_allclose(phasediff, 0, atol=0.1)
assert_allclose(amplitudediff, 0, atol=0.5)
def test_zscore(self):
for n in self.get_n():
x, y, xm, ym = self.generate_xy_sample(n)
#reference solution
zx = (x - x.mean()) / x.std()
zy = (y - y.mean()) / y.std()
#validate stats
assert_allclose(stats.zscore(x), zx, rtol=1e-10)
assert_allclose(stats.zscore(y), zy, rtol=1e-10)
#compare stats and mstats
assert_allclose(stats.zscore(x), stats.mstats.zscore(xm[0:len(x)]),
rtol=1e-10)
assert_allclose(stats.zscore(y), stats.mstats.zscore(ym[0:len(y)]),
rtol=1e-10)
def test_zscore(self):
for n in self.get_n():
x, y, xm, ym = self.generate_xy_sample(n)
#reference solution
zx = (x - x.mean()) / x.std()
zy = (y - y.mean()) / y.std()
#validate stats
assert_allclose(stats.zscore(x), zx, rtol=1e-10)
assert_allclose(stats.zscore(y), zy, rtol=1e-10)
#compare stats and mstats
assert_allclose(stats.zscore(x), stats.mstats.zscore(xm[0:len(x)]),
rtol=1e-10)
assert_allclose(stats.zscore(y), stats.mstats.zscore(ym[0:len(y)]),
rtol=1e-10)
def test_sem(self):
# example from stats.sem doc
a = np.arange(20).reshape(5, 4)
am = np.ma.array(a)
r = stats.sem(a, ddof=1)
rm = stats.mstats.sem(am, ddof=1)
assert_allclose(r, 2.82842712, atol=1e-5)
assert_allclose(rm, 2.82842712, atol=1e-5)
for n in self.get_n():
x, y, xm, ym = self.generate_xy_sample(n)
assert_almost_equal(stats.mstats.sem(xm, axis=None, ddof=0),
stats.sem(x, axis=None, ddof=0),
decimal=13)
assert_almost_equal(stats.mstats.sem(ym, axis=None, ddof=0),
stats.sem(y, axis=None, ddof=0),
decimal=13)
assert_almost_equal(stats.mstats.sem(xm, axis=None, ddof=1),
stats.sem(x, axis=None, ddof=1),
decimal=13)
assert_almost_equal(stats.mstats.sem(ym, axis=None, ddof=1),
stats.sem(y, axis=None, ddof=1),
decimal=13)
def test_perfect_locking_many_spiketrains_one_signal(self):
phases, amps, times = elephant.phase_analysis.spike_triggered_phase(
elephant.signal_processing.hilbert(self.anasig0),
[self.st0, self.st0],
interpolate=True)
assert_allclose(phases[0], -np.pi / 2.)
assert_allclose(amps[0], 1, atol=0.1)
assert_allclose(times[0].magnitude, self.st0.magnitude)
self.assertEqual(len(phases[0]), len(self.st0))
self.assertEqual(len(amps[0]), len(self.st0))
self.assertEqual(len(times[0]), len(self.st0))
def test_perfect_locking_many_spiketrains_one_signal(self):
phases, amps, times = elephant.phase_analysis.spike_triggered_phase(
elephant.signal_processing.hilbert(self.anasig0),
[self.st0, self.st0],
interpolate=True)
assert_allclose(phases[0], -np.pi / 2.)
assert_allclose(amps[0], 1, atol=0.1)
assert_allclose(times[0].magnitude, self.st0.magnitude)
self.assertEqual(len(phases[0]), len(self.st0))
self.assertEqual(len(amps[0]), len(self.st0))
self.assertEqual(len(times[0]), len(self.st0))
def test_vs_nonmasked(self):
np.random.seed(1234567)
outcome = np.random.randn(20, 4) + [0, 0, 1, 2]
# 1-D inputs
res1 = stats.ttest_1samp(outcome[:, 0], 1)
res2 = mstats.ttest_1samp(outcome[:, 0], 1)
assert_allclose(res1, res2)
# 2-D inputs
res1 = stats.ttest_1samp(outcome[:, 0], outcome[:, 1], axis=None)
res2 = mstats.ttest_1samp(outcome[:, 0], outcome[:, 1], axis=None)
assert_allclose(res1, res2)
res1 = stats.ttest_1samp(outcome[:, :2], outcome[:, 2:], axis=0)
res2 = mstats.ttest_1samp(outcome[:, :2], outcome[:, 2:], axis=0)
assert_allclose(res1, res2)
# Check default is axis=0
res3 = mstats.ttest_1samp(outcome[:, :2], outcome[:, 2:])
assert_allclose(res2, res3)
def test_vs_nonmasked(self):
np.random.seed(1234567)
outcome = np.random.randn(20, 4) + [0, 0, 1, 2]
# 1-D inputs
res1 = stats.ttest_1samp(outcome[:, 0], 1)
res2 = mstats.ttest_1samp(outcome[:, 0], 1)
assert_allclose(res1, res2)
# 2-D inputs
res1 = stats.ttest_1samp(outcome[:, 0], outcome[:, 1], axis=None)
res2 = mstats.ttest_1samp(outcome[:, 0], outcome[:, 1], axis=None)
assert_allclose(res1, res2)
res1 = stats.ttest_1samp(outcome[:, :2], outcome[:, 2:], axis=0)
res2 = mstats.ttest_1samp(outcome[:, :2], outcome[:, 2:], axis=0)
assert_allclose(res1, res2)
# Check default is axis=0
res3 = mstats.ttest_1samp(outcome[:, :2], outcome[:, 2:])
assert_allclose(res2, res3)
def test_rankdata(self):
for n in self.get_n():
x, y, xm, ym = self.generate_xy_sample(n)
r = stats.rankdata(x)
rm = stats.mstats.rankdata(x)
assert_allclose(r, rm)
def test_vs_nonmasked(self):
np.random.seed(1234567)
outcome = np.random.randn(20, 4) + [0, 0, 1, 2]
# 1-D inputs
res1 = stats.ttest_ind(outcome[:, 0], outcome[:, 1])
res2 = mstats.ttest_ind(outcome[:, 0], outcome[:, 1])
assert_allclose(res1, res2)
# 2-D inputs
res1 = stats.ttest_ind(outcome[:, 0], outcome[:, 1], axis=None)
res2 = mstats.ttest_ind(outcome[:, 0], outcome[:, 1], axis=None)
assert_allclose(res1, res2)
res1 = stats.ttest_ind(outcome[:, :2], outcome[:, 2:], axis=0)
res2 = mstats.ttest_ind(outcome[:, :2], outcome[:, 2:], axis=0)
assert_allclose(res1, res2)
# Check default is axis=0
res3 = mstats.ttest_ind(outcome[:, :2], outcome[:, 2:])
assert_allclose(res2, res3)
# Check equal_var
res4 = stats.ttest_ind(outcome[:, 0], outcome[:, 1], equal_var=True)
res5 = mstats.ttest_ind(outcome[:, 0], outcome[:, 1], equal_var=True)
assert_allclose(res4, res5)
res4 = stats.ttest_ind(outcome[:, 0], outcome[:, 1], equal_var=False)
res5 = mstats.ttest_ind(outcome[:, 0], outcome[:, 1], equal_var=False)
assert_allclose(res4, res5)
def test_rankdata(self):
for n in self.get_n():
x, y, xm, ym = self.generate_xy_sample(n)
r = stats.rankdata(x)
rm = stats.mstats.rankdata(x)
assert_allclose(r, rm)
def test_vs_nonmasked(self):
np.random.seed(1234567)
outcome = np.random.randn(20, 4) + [0, 0, 1, 2]
# 1-D inputs
res1 = stats.ttest_ind(outcome[:, 0], outcome[:, 1])
res2 = mstats.ttest_ind(outcome[:, 0], outcome[:, 1])
assert_allclose(res1, res2)
# 2-D inputs
res1 = stats.ttest_ind(outcome[:, 0], outcome[:, 1], axis=None)
res2 = mstats.ttest_ind(outcome[:, 0], outcome[:, 1], axis=None)
assert_allclose(res1, res2)
res1 = stats.ttest_ind(outcome[:, :2], outcome[:, 2:], axis=0)
res2 = mstats.ttest_ind(outcome[:, :2], outcome[:, 2:], axis=0)
assert_allclose(res1, res2)
# Check default is axis=0
res3 = mstats.ttest_ind(outcome[:, :2], outcome[:, 2:])
assert_allclose(res2, res3)
# Check equal_var
res4 = stats.ttest_ind(outcome[:, 0], outcome[:, 1], equal_var=True)
res5 = mstats.ttest_ind(outcome[:, 0], outcome[:, 1], equal_var=True)
assert_allclose(res4, res5)
res4 = stats.ttest_ind(outcome[:, 0], outcome[:, 1], equal_var=False)
res5 = mstats.ttest_ind(outcome[:, 0], outcome[:, 1], equal_var=False)
assert_allclose(res4, res5)
def test_linregress(self):
for n in self.get_n():
x, y, xm, ym = self.generate_xy_sample(n)
res1 = stats.linregress(x, y)
res2 = stats.mstats.linregress(xm, ym)
assert_allclose(np.asarray(res1), np.asarray(res2))
def test_zmap(self):
for n in self.get_n():
x, y, xm, ym = self.generate_xy_sample(n)
z = stats.zmap(x, y)
zm = stats.mstats.zmap(xm, ym)
assert_allclose(z, zm[0:len(z)], atol=1e-10)
def test_skewtest_2D_notmasked(self):
# a normal ndarray is passed to the masked function
x = np.random.random((20, 2)) * 20.
r = stats.skewtest(x)
rm = stats.mstats.skewtest(x)
assert_allclose(np.asarray(r), np.asarray(rm))
def test_zmap(self):
for n in self.get_n():
x, y, xm, ym = self.generate_xy_sample(n)
z = stats.zmap(x,y)
zm = stats.mstats.zmap(xm,ym)
assert_allclose(z, zm[0:len(z)], atol=1e-10)
def test_trimboth(self):
a = np.arange(20)
b = stats.trimboth(a, 0.1)
bm = stats.mstats.trimboth(a, 0.1)
assert_allclose(b, bm.data[~bm.mask])
def test_skewtest_2D_notmasked(self):
# a normal ndarray is passed to the masked function
x = np.random.random((20, 2)) * 20.
r = stats.skewtest(x)
rm = stats.mstats.skewtest(x)
assert_allclose(np.asarray(r), np.asarray(rm))
def test_trimboth(self):
a = np.arange(20)
b = stats.trimboth(a, 0.1)
bm = stats.mstats.trimboth(a, 0.1)
assert_allclose(np.sort(b), bm.data[~bm.mask])
def test_linregress(self):
for n in self.get_n():
x, y, xm, ym = self.generate_xy_sample(n)
res1 = stats.linregress(x, y)
res2 = stats.mstats.linregress(xm, ym)
assert_allclose(np.asarray(res1), np.asarray(res2))