def test_mad_std_with_axis():
data = np.array([[1, 2, 3, 4], [4, 3, 2, 1]])
# results follow data symmetry
result_axis0 = np.array([2.22390333, 0.74130111, 0.74130111, 2.22390333])
result_axis1 = np.array([1.48260222, 1.48260222])
assert_allclose(funcs.mad_std(data, axis=0), result_axis0)
assert_allclose(funcs.mad_std(data, axis=1), result_axis1)
def test_mad_std_with_axis_and_nan():
data = np.array([[1, 2, 3, 4, np.nan], [4, 3, 2, 1, np.nan]])
# results follow data symmetry
result_axis0 = np.array(
[2.22390333, 0.74130111, 0.74130111, 2.22390333, np.nan])
result_axis1 = np.array([1.48260222, 1.48260222])
assert_allclose(funcs.mad_std(data, axis=0, ignore_nan=True), result_axis0)
assert_allclose(funcs.mad_std(data, axis=1, ignore_nan=True), result_axis1)
def test_mad_std_with_axis():
data = np.array([[1, 2, 3, 4],
[4, 3, 2, 1]])
# results follow data symmetry
result_axis0 = np.array([2.22390333, 0.74130111, 0.74130111,
2.22390333])
result_axis1 = np.array([1.48260222, 1.48260222])
assert_allclose(funcs.mad_std(data, axis=0), result_axis0)
assert_allclose(funcs.mad_std(data, axis=1), result_axis1)
def test_mad_std_with_axis_and_nan_array_type():
# mad_std should return a masked array if given one, and not otherwise
data = np.array([[1, 2, 3, 4, np.nan], [4, 3, 2, 1, np.nan]])
result = funcs.mad_std(data, axis=0, ignore_nan=True)
assert not np.ma.isMaskedArray(result)
data = np.ma.masked_where(np.isnan(data), data)
result = funcs.mad_std(data, axis=0, ignore_nan=True)
assert np.ma.isMaskedArray(result)
def test_mad_std_with_axis_and_nan():
data = np.array([[1, 2, 3, 4, np.nan],
[4, 3, 2, 1, np.nan]])
# results follow data symmetry
result_axis0 = np.array([2.22390333, 0.74130111, 0.74130111,
2.22390333, np.nan])
result_axis1 = np.array([1.48260222, 1.48260222])
assert_allclose(funcs.mad_std(data, axis=0, ignore_nan=True), result_axis0)
assert_allclose(funcs.mad_std(data, axis=1, ignore_nan=True), result_axis1)
def test_mad_std_withnan():
with NumpyRNGContext(12345):
data = np.empty([102, 102])
data[:] = np.nan
data[1:-1, 1:-1] = np.random.normal(5, 2, size=(100, 100))
assert_allclose(funcs.mad_std(data, ignore_nan=True), 2.0, rtol=0.05)
assert np.isnan(funcs.mad_std([1, 2, 3, 4, 5, np.nan]))
assert_allclose(funcs.mad_std([1, 2, 3, 4, 5, np.nan], ignore_nan=True),
1.482602218505602)
def test_mad_std_withnan():
with NumpyRNGContext(12345):
data = np.empty([102, 102])
data[:] = np.nan
data[1:-1, 1:-1] = np.random.normal(5, 2, size=(100, 100))
assert_allclose(funcs.mad_std(data, ignore_nan=True), 2.0, rtol=0.05)
assert np.isnan(funcs.mad_std([1, 2, 3, 4, 5, np.nan]))
assert_allclose(funcs.mad_std([1, 2, 3, 4, 5, np.nan], ignore_nan=True),
1.482602218505602)
def test_mad_std_with_axis_and_nan_array_type():
# mad_std should return a masked array if given one, and not otherwise
data = np.array([[1, 2, 3, 4, np.nan], [4, 3, 2, 1, np.nan]])
with pytest.warns(RuntimeWarning, match=r'All-NaN slice encountered'):
result = funcs.mad_std(data, axis=0, ignore_nan=True)
assert not np.ma.isMaskedArray(result)
data = np.ma.masked_where(np.isnan(data), data)
result = funcs.mad_std(data, axis=0, ignore_nan=True)
assert np.ma.isMaskedArray(result)
def test_mad_std_with_axis_and_nan_array_type():
# mad_std should return a masked array if given one, and not otherwise
data = np.array([[1, 2, 3, 4, np.nan],
[4, 3, 2, 1, np.nan]])
result = funcs.mad_std(data, axis=0, ignore_nan=True)
assert not np.ma.isMaskedArray(result)
data = np.ma.masked_where(np.isnan(data), data)
result = funcs.mad_std(data, axis=0, ignore_nan=True)
assert np.ma.isMaskedArray(result)
def test_mad_std_with_axis_and_nan():
data = np.array([[1, 2, 3, 4, np.nan],
[4, 3, 2, 1, np.nan]])
# results follow data symmetry
result_axis0 = np.array([2.22390333, 0.74130111, 0.74130111,
2.22390333, np.nan])
result_axis1 = np.array([1.48260222, 1.48260222])
with pytest.warns(RuntimeWarning,
match=r'All-NaN slice encountered'):
assert_allclose(funcs.mad_std(data, axis=0, ignore_nan=True), result_axis0)
assert_allclose(funcs.mad_std(data, axis=1, ignore_nan=True), result_axis1)
def test_mad_std_scalar_return():
with NumpyRNGContext(12345):
data = np.random.normal(5, 2, size=(10, 10))
# make a masked array with no masked points
data = np.ma.masked_where(np.isnan(data), data)
rslt = funcs.mad_std(data)
# want a scalar result, NOT a masked array
assert np.isscalar(rslt)
data[5, 5] = np.nan
rslt = funcs.mad_std(data, ignore_nan=True)
assert np.isscalar(rslt)
with catch_warnings():
rslt = funcs.mad_std(data)
assert np.isscalar(rslt)
assert np.isnan(rslt)
def test_mad_std_warns():
with NumpyRNGContext(12345):
data = np.random.normal(5, 2, size=(10, 10))
data[5, 5] = np.nan
with catch_warnings() as warns:
rslt = funcs.mad_std(data, ignore_nan=False)
assert np.isnan(rslt)
def test_mad_std_warns():
with NumpyRNGContext(12345):
data = np.random.normal(5, 2, size=(10, 10))
data[5, 5] = np.nan
with catch_warnings() as warns:
rslt = funcs.mad_std(data, ignore_nan=False)
assert np.isnan(rslt)
def test_mad_std_scalar_return():
with NumpyRNGContext(12345):
data = np.random.normal(5, 2, size=(10, 10))
# make a masked array with no masked points
data = np.ma.masked_where(np.isnan(data), data)
rslt = funcs.mad_std(data)
# want a scalar result, NOT a masked array
assert np.isscalar(rslt)
data[5, 5] = np.nan
rslt = funcs.mad_std(data, ignore_nan=True)
assert np.isscalar(rslt)
with catch_warnings():
rslt = funcs.mad_std(data)
assert np.isscalar(rslt)
try:
assert not np.isnan(rslt)
# This might not be an issue anymore when only numpy>=1.13 is
# supported. NUMPY_LT_1_13 xref #7267
except AssertionError:
pytest.xfail('See #5232')
def test_mad_std_scalar_return():
with NumpyRNGContext(12345):
data = np.random.normal(5, 2, size=(10, 10))
# make a masked array with no masked points
data = np.ma.masked_where(np.isnan(data), data)
rslt = funcs.mad_std(data)
# want a scalar result, NOT a masked array
assert np.isscalar(rslt)
data[5, 5] = np.nan
rslt = funcs.mad_std(data, ignore_nan=True)
assert np.isscalar(rslt)
with catch_warnings():
rslt = funcs.mad_std(data)
assert np.isscalar(rslt)
try:
assert not np.isnan(rslt)
# This might not be an issue anymore when only numpy>=1.13 is
# supported. NUMPY_LT_1_13 xref #7267
except AssertionError:
pytest.xfail('See #5232')
def test_mad_std_warns():
if NUMPY_LT_1_17:
ctx = pytest.warns(RuntimeWarning,
match='Invalid value encountered in median')
else:
ctx = nullcontext()
with NumpyRNGContext(12345):
data = np.random.normal(5, 2, size=(10, 10))
data[5, 5] = np.nan
with ctx:
rslt = funcs.mad_std(data, ignore_nan=False)
assert np.isnan(rslt)
def test_mad_std_scalar_return():
with NumpyRNGContext(12345):
data = np.random.normal(5, 2, size=(10, 10))
# make a masked array with no masked points
data = np.ma.masked_where(np.isnan(data), data)
rslt = funcs.mad_std(data)
# want a scalar result, NOT a masked array
assert np.isscalar(rslt)
if NUMPY_LT_1_17:
ctx = pytest.warns(RuntimeWarning,
match='Invalid value encountered in median')
else:
ctx = nullcontext()
data[5, 5] = np.nan
with ctx:
rslt = funcs.mad_std(data, ignore_nan=True)
assert np.isscalar(rslt)
rslt = funcs.mad_std(data)
assert np.isscalar(rslt)
assert np.isnan(rslt)
def _nanmadstd(array, axis=None):
"""mad_std function that ignores NaNs by default."""
return mad_std(array, axis=axis, ignore_nan=True)
def test_mad_std():
with NumpyRNGContext(12345):
data = np.random.normal(5, 2, size=(100, 100))
assert_allclose(funcs.mad_std(data), 2.0, rtol=0.05)
def test_mad_std():
with NumpyRNGContext(12345):
data = np.random.normal(5, 2, size=(100, 100))
assert_allclose(funcs.mad_std(data), 2.0, rtol=0.05)