def test_2d_waxis(self):
# Tests median w/ 2D arrays and different axis.
x = masked_array(np.arange(30).reshape(10, 3))
x[:3] = x[-3:] = masked
assert_equal(median(x), 14.5)
assert_equal(median(x, axis=0), [13.5, 14.5, 15.5])
assert_equal(median(x, axis=1), [0, 0, 0, 10, 13, 16, 19, 0, 0, 0])
assert_equal(median(x, axis=1).mask, [1, 1, 1, 0, 0, 0, 0, 1, 1, 1])
def test_2d_waxis(self):
# Tests median w/ 2D arrays and different axis.
x = masked_array(np.arange(30).reshape(10, 3))
x[:3] = x[-3:] = masked
assert_equal(median(x), 14.5)
assert_equal(median(x, axis=0), [13.5, 14.5, 15.5])
assert_equal(median(x, axis=1), [0, 0, 0, 10, 13, 16, 19, 0, 0, 0])
assert_equal(median(x, axis=1).mask, [1, 1, 1, 0, 0, 0, 0, 1, 1, 1])
def test_3d(self):
# Tests median w/ 3D
x = np.ma.arange(24).reshape(3, 4, 2)
x[x % 3 == 0] = masked
assert_equal(median(x, 0), [[12, 9], [6, 15], [12, 9], [18, 15]])
x.shape = (4, 3, 2)
assert_equal(median(x, 0), [[99, 10], [11, 99], [13, 14]])
x = np.ma.arange(24).reshape(4, 3, 2)
x[x % 5 == 0] = masked
assert_equal(median(x, 0), [[12, 10], [8, 9], [16, 17]])
def test_3d(self):
# Tests median w/ 3D
x = np.ma.arange(24).reshape(3, 4, 2)
x[x % 3 == 0] = masked
assert_equal(median(x, 0), [[12, 9], [6, 15], [12, 9], [18, 15]])
x.shape = (4, 3, 2)
assert_equal(median(x, 0), [[99, 10], [11, 99], [13, 14]])
x = np.ma.arange(24).reshape(4, 3, 2)
x[x % 5 == 0] = masked
assert_equal(median(x, 0), [[12, 10], [8, 9], [16, 17]])
def test_2d(self):
# Tests median w/ 2D
(n, p) = (101, 30)
x = masked_array(np.linspace(-1., 1., n), )
x[:10] = x[-10:] = masked
z = masked_array(np.empty((n, p), dtype=float))
z[:, 0] = x[:]
idx = np.arange(len(x))
for i in range(1, p):
np.random.shuffle(idx)
z[:, i] = x[idx]
assert_equal(median(z[:, 0]), 0)
assert_equal(median(z), 0)
assert_equal(median(z, axis=0), np.zeros(p))
assert_equal(median(z.T, axis=1), np.zeros(p))
def test_2d(self):
# Tests median w/ 2D
(n, p) = (101, 30)
x = masked_array(np.linspace(-1., 1., n),)
x[:10] = x[-10:] = masked
z = masked_array(np.empty((n, p), dtype=float))
z[:, 0] = x[:]
idx = np.arange(len(x))
for i in range(1, p):
np.random.shuffle(idx)
z[:, i] = x[idx]
assert_equal(median(z[:, 0]), 0)
assert_equal(median(z), 0)
assert_equal(median(z, axis=0), np.zeros(p))
assert_equal(median(z.T, axis=1), np.zeros(p))
def test_out(self):
x = masked_array(np.arange(30).reshape(10, 3))
x[:3] = x[-3:] = masked
out = masked_array(np.ones(10))
r = median(x, axis=1, out=out)
assert_equal(r, out)
assert_(type(r) == MaskedArray)
def test_out(self):
x = masked_array(np.arange(30).reshape(10, 3))
x[:3] = x[-3:] = masked
out = masked_array(np.ones(10))
r = median(x, axis=1, out=out)
assert_equal(r, out)
assert_(type(r) == MaskedArray)
def test_neg_axis(self):
x = masked_array(np.arange(30).reshape(10, 3))
x[:3] = x[-3:] = masked
assert_equal(median(x, axis=-1), median(x, axis=1))
def test_neg_axis(self):
x = masked_array(np.arange(30).reshape(10, 3))
x[:3] = x[-3:] = masked
assert_equal(median(x, axis=-1), median(x, axis=1))
def get_sigma_median_heuristic(X):
dists=squareform(pdist(X, 'euclidean'))
median_dist=median(dists[dists>0])
sigma=sqrt(0.5*median_dist)
return sigma
