def test_mad(self):
"""Test gesd on almost flat data."""
number_outliers, suspicious_indexes, test_statistics, critical_values, all_z_scores = gesd(
self.series, mad=True)
assert 0 == number_outliers
assert [] == suspicious_indexes
assert [] == test_statistics
assert [] == critical_values
assert [] == all_z_scores
def test_flat(self):
"""Test gesd on flat data."""
number_outliers, suspicious_indexes, test_statistics, critical_values, all_z_scores = gesd(
[1] * 20)
assert 0 == number_outliers
assert [] == suspicious_indexes
assert [] == test_statistics
assert [] == critical_values
assert [] == all_z_scores
def test_standard(self):
"""Test gesd on almost flat data."""
number_outliers, suspicious_indexes, _, _, _ = gesd(self.series)
assert 1 == number_outliers
assert [203] == suspicious_indexes
def test_mad(self):
""" Test MAD z score. """
number_outliers, suspicious_indexes, test_statistics, critical_values, all_z_scores = gesd(
self.ROSNER, mad=True)
assert 4 == number_outliers
assert np.array_equal(suspicious_indexes[:number_outliers],
self.CANONICAL_INDEXES[:number_outliers])
def test_significance_level_gt_one(self):
"""Test invalid significance_level."""
with pytest.raises(ValueError, match=r"invalid significance_level"):
gesd([1] * 20, significance_level=1)
def test_insufficient_data(self):
"""Test insufficient data."""
with pytest.raises(ValueError, match=r"max_outliers.* <= length"):
gesd([1])
with pytest.raises(ValueError, match=r"max_outliers.* <= length"):
gesd([1] * 10)
def test_canonical(self):
"""Test gesd implementation."""
number_outliers, suspicious_indexes, test_statistics, critical_values, all_z_scores = gesd(
self.ROSNER)
assert 3 == number_outliers
assert np.array_equal(suspicious_indexes, self.CANONICAL_INDEXES)
assert all(
np.isclose(self.CANONICAL_STATS,
np.fabs(test_statistics),
rtol=0.001))
assert all(
np.isclose(self.CANONICAL_CRITICAL, critical_values, rtol=0.001))
def test_max_outliers(self):
"""Test insufficient data."""
with pytest.raises(ValueError, match=r"max_outliers.* must be >= 1"):
gesd([1], 0)
with pytest.raises(ValueError, match=r"max_outliers.* must be >= 1"):
gesd([1], -1)
def test_triple_max_10(self):
"""Test gesd on flat data."""
number_outliers, suspicious_indexes, test_statistics, critical_values, all_z_scores = gesd(
TRIPLE, mad=True)
assert 3 == number_outliers
assert [7, 20, 30] == suspicious_indexes[:3]
def test_single_max_10(self):
"""Test gesd on flat data."""
number_outliers, suspicious_indexes, test_statistics, critical_values, all_z_scores = gesd(
SINGLE, mad=True)
assert 10 == number_outliers
assert 7 == suspicious_indexes[0]
def test_double_max_10(self):
"""Test gesd on flat data."""
number_outliers, suspicious_indexes, test_statistics, critical_values, all_z_scores = gesd(
DOUBLE, max_outliers=2, mad=True)
assert 2 == number_outliers
assert [7, 20] == suspicious_indexes
def test_double_max_10(self):
"""Test gesd on flat data."""
number_outliers, suspicious_indexes, test_statistics, critical_values, all_z_scores = gesd(
DOUBLE)
assert 2 == number_outliers
assert 7 == suspicious_indexes[0]
assert 20 == suspicious_indexes[1]
def test_no_data(self):
"""Test no data."""
with pytest.raises(ValueError, match=r"No Data"):
gesd(None)
with pytest.raises(ValueError, match=r"No Data"):
gesd([])
def test_single(self):
"""Test gesd on flat data."""
number_outliers, suspicious_indexes, test_statistics, critical_values, all_z_scores = gesd(
SINGLE, max_outliers=1)
assert 1 == number_outliers
assert [7] == suspicious_indexes
def test_mad(self, mad_series):
"""Test gesd on real data with Median Absolute Deviation."""
number_outliers, suspicious_indexes, _, _, _ = gesd(mad_series,
mad=True)
assert 2 == number_outliers
assert [90, 4, 27, 37, 64, 78, 74, 88, 47, 60] == suspicious_indexes
def test_standard(self, real_series):
"""Test gesd on real data with standard."""
number_outliers, suspicious_indexes, _, _, _ = gesd(real_series)
assert 2 == number_outliers
assert [90, 4, 27, 88, 37, 64, 126, 78, 74, 47] == suspicious_indexes
