def test_sw_detect_multi(self):
"""Test sw_detect_multi"""
data_full = file_full.get('data')
sw = sw_detect_multi(data_full, sf_full, chan_full)
sw_no_out = sw_detect_multi(data_full, sf_full, chan_full,
remove_outliers=True)
assert sw_no_out.shape[0] < sw.shape[0]
bv = get_bool_vector(data_full, sf_full, sw)
assert bv.shape[0] == len(chan_full)
# Test with hypnogram
sw_detect_multi(data_full, sf_full, chan_full,
hypno=hypno_full, include=3)
# Now we replace one channel with no slow-wave / bad data
data_full[1, :] = np.random.random(data_full.shape[1])
# Test where data.shape[0] != len(chan)
with pytest.raises(AssertionError):
sw_detect_multi(data_full, sf_full, chan_full[:-1])
# Test with only bad channels
data_full[0, :] = np.random.random(data_full.shape[1])
data_full[2, :] = np.random.random(data_full.shape[1])
with self.assertLogs('yasa', level='WARNING'):
sp = sw_detect_multi(data_full, sf_full, chan_full)
assert sp is None
def test_spindles_detect_multi(self):
"""Test spindles_detect_multi"""
sp = spindles_detect_multi(data_full, sf_full, chan_full)
sp_no_out = spindles_detect_multi(data_full, sf_full, chan_full,
remove_outliers=True)
sp_multi = spindles_detect_multi(data_full, sf_full, chan_full,
multi_only=True)
assert sp_multi.shape[0] < sp.shape[0]
assert sp_no_out.shape[0] < sp.shape[0]
bv = get_bool_vector(data_full, sf_full, sp)
assert bv.shape[0] == len(chan_full)
# Test with hypnogram
spindles_detect_multi(data_full, sf_full, chan_full, hypno=hypno_full,
include=2)
# Now we replace one channel with no spindle / bad data
data_full[1, :] = np.random.random(data_full.shape[1])
# Test where data.shape[0] != len(chan)
with pytest.raises(AssertionError):
spindles_detect_multi(data_full, sf_full, chan_full[:-1])
# Test with only bad channels
data_full[0, :] = np.random.random(data_full.shape[1])
data_full[2, :] = np.random.random(data_full.shape[1])
with self.assertLogs('yasa', level='WARNING'):
sp = spindles_detect_multi(data_full, sf_full, chan_full)
assert sp is None
def test_rem_detect(self):
"""Test function REM detect
"""
file_rem = np.load('notebooks/data_EOGs_REM_256Hz.npz')
data_rem = file_rem['data']
loc, roc = data_rem[0, :], data_rem[1, :]
sf_rem = file_rem['sf']
# chan_rem = file_rem['chan']
hypno_rem = 4 * np.ones_like(loc)
# Parameters product testing
freq_rem = [(0.5, 5), (0.3, 8)]
duration = [(0.3, 1.5), [0.5, 1]]
amplitude = [(50, 200), [60, 300]]
downsample = [True, False]
hypno = [hypno_rem, None]
prod_args = product(freq_rem, duration, amplitude, downsample, hypno)
for i, (f, dr, am, ds, h) in enumerate(prod_args):
rem_detect(loc,
roc,
sf_rem,
hypno=h,
freq_rem=f,
duration=dr,
amplitude=am,
downsample=ds)
# With isolation forest
df_rem = rem_detect(loc, roc, sf)
df_rem2 = rem_detect(loc, roc, sf, remove_outliers=True)
assert df_rem.shape[0] > df_rem2.shape[0]
assert get_bool_vector(loc, sf, df_rem).size == loc.size
# With REM hypnogram
hypno_rem = 4 * np.ones_like(loc)
df_rem = rem_detect(loc, roc, sf_full, hypno=hypno_rem)
hypno_rem = np.r_[np.ones(int(loc.size / 2)),
4 * np.ones(int(loc.size / 2))]
df_rem2 = rem_detect(loc, roc, sf_full, hypno=hypno_rem)
assert df_rem.shape[0] > df_rem2.shape[0]
# Test with wrong data amplitude on ROC
with self.assertLogs('yasa', level='ERROR'):
rd = rem_detect(loc * 1e-8, roc, sf)
assert rd is None
# Test with wrong data amplitude on LOC
with self.assertLogs('yasa', level='ERROR'):
rd = rem_detect(loc, roc * 1e8, sf)
assert rd is None
# Hypnogram with sf = 150
with self.assertLogs('yasa', level='WARNING'):
rem_detect(loc, roc * 1e8, sf=150) # Fake sampling frequency
# No values in hypno intersect with include
with self.assertLogs('yasa', level='ERROR'):
sp = rem_detect(loc, roc, sf_full, hypno=hypno_rem, include=5)
assert sp is None
def test_get_bool_vector(self):
"""Test functions get_bool_vector"""
# Numpy
sp_params = spindles_detect(data, sf)
out = get_bool_vector(data, sf, sp_params)
assert out.size == data.size
assert np.unique(out).size == 2
# MNE
# Single channel
sp_params = spindles_detect_multi(data_mne_single)
out = get_bool_vector(data_mne_single, detection=sp_params)
assert out.size == max(data_mne_single.get_data().shape)
assert np.unique(out).size == 2
# Multi-channels
sp_params = spindles_detect_multi(data_mne)
out = get_bool_vector(data_mne, detection=sp_params)
assert out.size == data_mne.get_data().size
assert np.unique(out).size == 2
def test_get_bool_vector(self):
"""Test functions get_bool_vector"""
sp_params = spindles_detect(data, sf)
out = get_bool_vector(data, sf, sp_params)
assert out.size == data.size
assert np.unique(out).size == 2
