def test_focalmax(self):
np.random.seed(0)
a = np.random.rand(100, 100)
self.assertTrue(
np.allclose(
rolling_window(a, 3).max(axis=(2, 3)),
focal_statistics(a, window_size=3, func="max")[1:-1, 1:-1]))
def test_focalnanmax_reduce(self):
np.random.seed(0)
a = np.random.rand(5, 5)
max = a.max()
a[1, 1] = np.nan
self.assertTrue(
focal_statistics(a, window_size=5, func="max", reduce=True) == max)
def test_focalnanmin(self):
np.random.seed(0)
a = np.random.rand(5, 5)
min = a.min()
a[1, 1] = np.nan
self.assertTrue(
focal_statistics(a, window_size=5, func="min")[2, 2] == min)
def test_focalmean_reduce(self):
np.random.seed(0)
a = np.random.rand(100, 100)
self.assertTrue(
np.allclose(
rolling_window(a, 5, reduce=True).mean(axis=(2, 3)),
focal_statistics(a, window_size=5, func="mean", reduce=True)))
def test_focalstd_df_1(self):
np.random.seed(0)
a = np.random.rand(100, 100)
self.assertTrue(
np.allclose(
rolling_window(a, 3).std(axis=(2, 3), ddof=1),
focal_statistics(a, window_size=3, func="std",
std_df=1)[1:-1, 1:-1]))
def test_errors(self):
np.random.seed(0)
a = np.random.rand(100, 100)
with self.assertRaises(IndexError):
focal_statistics(a, window_size=101, func="max")
with self.assertRaises(ValueError):
focal_statistics(a,
window_size=3,
fraction_accepted=1.1,
func="max")
with self.assertRaises(TypeError):
focal_statistics(a, window_size="a", func="max")
with self.assertRaises(ValueError):
focal_statistics(np.random.rand(100, 100, 100),
window_size=3,
func="max")
with self.assertRaises(ValueError):
focal_statistics(a, window_size=9, reduce=True, func="max")
def test_focalmean_reduce_fraction_accepted_1(self):
np.random.seed(0)
a = np.random.rand(5, 5)
# return value when all values are present
self.assertTrue(
np.allclose(
focal_statistics(a,
window_size=5,
func="mean",
reduce=True,
fraction_accepted=1),
np.mean(rolling_window(a, 5).mean(axis=(2, 3)))))
a[3, 3] = np.nan
# return nan when only one value is missing
self.assertTrue(
np.allclose(focal_statistics(a,
window_size=5,
func="mean",
reduce=True,
fraction_accepted=1),
np.array([[np.nan]]),
equal_nan=True))
def test_focalmean_reduce_fraction_accepted_0(self):
np.random.seed(0)
a = np.full((5, 5), np.nan)
# return nan when nothing is present
self.assertTrue(
np.allclose(focal_statistics(a,
window_size=5,
func="mean",
reduce=True,
fraction_accepted=0),
np.array([[np.nan]]),
equal_nan=True))
# return value if only one value is present
a[3, 3] = 10
self.assertTrue(
np.allclose(
focal_statistics(a,
window_size=5,
func="mean",
reduce=True,
fraction_accepted=0),
np.nanmean(rolling_window(a, 5, reduce=True))))
