def test_haversine_decomposition():
# Distancia a-->c <= a-->b + b-->c (metric='haversine')
p_a = [1, 1]
p_b = [2, 2]
p_c = [1, 2]
dist_ab = distances.haversine(np.array(p_a), np.array([p_b]), 2)
dist_ac = distances.haversine(np.array(p_a), np.array([p_c]), 2)
dist_bc = distances.haversine(np.array(p_b), np.array([p_c]), 2)
assert dist_ac <= dist_ab + dist_bc
def test_haversine_simetric():
# Distancia a-->b = b-->a (metric='haversine')
dist1 = distances.haversine(c0=np.array([1, 1]),
centres=np.array([[2, 2]]),
dim=2)
dist2 = distances.haversine(c0=np.array([2, 2]),
centres=np.array([[1, 1]]),
dim=2)
np.testing.assert_almost_equal(dist1, dist2, decimal=10)
def test_haversine_not_nan():
# Distancias != NaN (metric='haversine')
random = np.random.RandomState(42)
data = random.uniform(-10, 10, size=(10, 2))
centre_0 = random.uniform(-10, 10, size=(2, ))
dist = distances.haversine(centre_0, data, 2)
assert not np.isnan(dist).any()
def test_haversine_gt0():
# Distancias >= 0 (metric='haversine')
random = np.random.RandomState(42)
data = random.uniform(-10, 10, size=(10, 2))
centre_0 = random.uniform(-10, 10, size=(2, ))
dist = distances.haversine(centre_0, data, 2)
assert (dist >= 0).all()