def ksc_runner(self, method):
k = 2
#One cluster with uniform series, another with a peak.
X = np.array([[1.0, 1, 1], [1.1, 1, 1], [1.2, 1, 1], [1.3, 1, 1],
[1.3, 1, 1], [1.3, 1, 1], [1.3, 1, 1], [1.3, 1, 1],
[90, 2000, 90], [90, 2001, 90], [90, 2002, 90],
[90, 2003, 90]])
cents, assign, shift, distc = ksc.ksc(X, k)
del shift
self.assertEqual(len(set(assign)), k)
self.assertEqual(sum(assign == assign[0]), 8)
self.assertEqual(sum(assign == assign[-1]), 4)
self.assertEqual(len(set(assign[:8])), 1)
self.assertEqual(len(set(assign[8:])), 1)
self.assertFalse(set(assign[:8]) == set(assign[8:]))
cluster_one = assign[0]
cluster_two = assign[-1]
self.assertTrue(dist.dist(X[0], cents[cluster_one]) < \
dist.dist(X[0], cents[cluster_two]))
self.assertTrue(dist.dist(cents[cluster_one], cents[cluster_two]) > 0)
for i in xrange(X.shape[0]):
self.assertAlmostEqual(dist.dist(X[i], cents[0], True),
distc.T[i, 0], 5)
self.assertAlmostEqual(dist.dist(X[i], cents[1], True),
distc.T[i, 1], 5)
def test_dist(self):
array1 = np.array([])
array2 = np.array([])
self.assertEqual(0, dist.dist(array1, array2))
array1 = np.array([0.0, 0.0])
array2 = np.array([0.0, 0.0])
self.assertEqual(0, dist.dist(array1, array2))
array1 = np.array([1.0, 2.0])
array2 = np.array([0.0, 0.0])
self.assertEqual(1, dist.dist(array1, array2))
array1 = np.array([0.0, 0.0])
array2 = np.array([1.0, 2.0])
self.assertEqual(1, dist.dist(array1, array2))
array1 = np.array([2.0, 3.0, 4.0])
array2 = np.array([3.0, 4.0, 0.0])
self.assertAlmostEqual(0, dist.dist(array1, array1))
self.assertAlmostEqual(0, dist.dist(array2, array2))
expected = 2 / sqrt(29)
self.assertAlmostEqual(expected, dist.dist(array1, array2, True))
self.assertAlmostEqual(expected, dist.dist(array2, array1, True))
def test_dist(self):
array1 = np.array([])
array2 = np.array([])
self.assertEqual(0, dist.dist(array1, array2))
array1 = np.array([0., 0.])
array2 = np.array([0., 0.])
self.assertEqual(0, dist.dist(array1, array2))
array1 = np.array([1., 2.])
array2 = np.array([0., 0.])
self.assertEqual(1, dist.dist(array1, array2))
array1 = np.array([0., 0.])
array2 = np.array([1., 2.])
self.assertEqual(1, dist.dist(array1, array2))
array1 = np.array([2.0, 3.0, 4.0])
array2 = np.array([3.0, 4.0, 0.0])
self.assertAlmostEqual(0, dist.dist(array1, array1))
self.assertAlmostEqual(0, dist.dist(array2, array2))
expected = 2 / sqrt(29)
self.assertAlmostEqual(expected, dist.dist(array1, array2, True))
self.assertAlmostEqual(expected, dist.dist(array2, array1, True))
def ksc_runner(self, method):
k = 2
# One cluster with uniform series, another with a peak.
X = np.array(
[
[1.0, 1, 1],
[1.1, 1, 1],
[1.2, 1, 1],
[1.3, 1, 1],
[1.3, 1, 1],
[1.3, 1, 1],
[1.3, 1, 1],
[1.3, 1, 1],
[90, 2000, 90],
[90, 2001, 90],
[90, 2002, 90],
[90, 2003, 90],
]
)
cents, assign, shift, distc = ksc.ksc(X, k)
del shift
self.assertEqual(len(set(assign)), k)
self.assertEqual(sum(assign == assign[0]), 8)
self.assertEqual(sum(assign == assign[-1]), 4)
self.assertEqual(len(set(assign[:8])), 1)
self.assertEqual(len(set(assign[8:])), 1)
self.assertFalse(set(assign[:8]) == set(assign[8:]))
cluster_one = assign[0]
cluster_two = assign[-1]
self.assertTrue(dist.dist(X[0], cents[cluster_one]) < dist.dist(X[0], cents[cluster_two]))
self.assertTrue(dist.dist(cents[cluster_one], cents[cluster_two]) > 0)
for i in xrange(X.shape[0]):
self.assertAlmostEqual(dist.dist(X[i], cents[0], True), distc.T[i, 0], 5)
self.assertAlmostEqual(dist.dist(X[i], cents[1], True), distc.T[i, 1], 5)
