def test_matching_dissim(self):
a = np.array([[0, 1, 2, 0, 1, 2]])
b = np.array([[0, 1, 2, 0, 1, 0]])
assert_equal(1, matching_dissim(a, b))
a = np.array([[np.NaN, 1, 2, 0, 1, 2]])
b = np.array([[0, 1, 2, 0, 1, 0]])
assert_equal(2, matching_dissim(a, b))
a = np.array([['a', 'b', 'c', 'd']])
b = np.array([['a', 'b', 'c', 'd'], ['d', 'c', 'b', 'a']])
assert_array_equal(np.array([0, 4]), matching_dissim(a, b))
def test_matching_dissim(self):
a = np.array([[0, 1, 2, 0, 1, 2]])
b = np.array([[0, 1, 2, 0, 1, 0]])
assert_equal(1, matching_dissim(a, b))
a = np.array([[np.NaN, 1, 2, 0, 1, 2]])
b = np.array([[0, 1, 2, 0, 1, 0]])
assert_equal(2, matching_dissim(a, b))
a = np.array([['a', 'b', 'c', 'd']])
b = np.array([['a', 'b', 'c', 'd'], ['d', 'c', 'b', 'a']])
assert_array_equal(np.array([0, 4]), matching_dissim(a, b))
def _get_initial_centers(self, dataset, categorical_indices):
dataset_cat = dataset.take(categorical_indices, axis=1).values
categorical_labels = [
column for index, column in enumerate(dataset.columns)
if index in categorical_indices
]
dataset_num = dataset.drop(categorical_labels, axis=1).values
categorical_weight = self.categorical_weight
if categorical_weight is None or categorical_weight < 0:
categorical_weight = 0.5 * dataset_num.std()
initial_centroids_num = np.zeros(
(self.cluster_number, dataset_num.shape[1]))
initial_centroids_cat = np.zeros(
(self.cluster_number, dataset_cat.shape[1]))
rand_index = randint(0, dataset.shape[0] - 1)
initial_centroids_num[0], initial_centroids_cat[0] = dataset_num[
rand_index], dataset_cat[rand_index]
for i in range(1, self.cluster_number):
distances_num_cat = [
np.zeros((i, dataset.shape[0]), dtype=np.float64),
np.zeros((i, dataset.shape[0]))
]
for j in range(0, i):
distances_num_cat[0][j] = dissimilarity_python.euclidean(
dataset_num, initial_centroids_num[j])
distances_num_cat[1][j] = matching_dissim(
dataset_cat, initial_centroids_cat[j])
distances = np.amin(distances_num_cat[0] +
categorical_weight * distances_num_cat[1],
axis=0)
probabilities = distances / np.sum(distances)
chosen_point = np.random.choice(range(0, dataset.shape[0]),
p=probabilities)
initial_centroids_num[i] = dataset_num[chosen_point]
initial_centroids_cat[i] = dataset_cat[chosen_point]
initial_centroids = [initial_centroids_num, initial_centroids_cat]
return initial_centroids
def test_ng_dissim(self):
X = np.array([[0, 1, 2, 0, 1, 2], [0, 1, 2, 0, 1, 1]])
centroids = X
membship = np.array([[1, 0], [0, 1]])
assert_array_equal(np.array([0., 1.]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([1., 0.]),
ng_dissim(centroids, X[1], X=X, membship=membship))
# Unit test for initialization (i.e., same as matching_dissim)
membship = np.array([[0, 0], [0, 0]])
mdiss_00 = matching_dissim(np.array([X[0]]), np.array([X[0]]))[0]
mdiss_01 = matching_dissim(np.array([X[0]]), np.array([X[1]]))[0]
mdiss_11 = matching_dissim(np.array([X[1]]), np.array([X[1]]))[0]
assert_array_equal(np.array([mdiss_00, mdiss_01]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([mdiss_01, mdiss_11]),
ng_dissim(centroids, X[1], X=X, membship=membship))
# Unit test for NaN
X = np.array([[np.NaN, 1, 2, 0, 1, 2], [0, 1, 2, 0, 1, 1]])
centroids = X
membship = np.array([[1, 0], [0, 1]])
assert_array_equal(np.array([1., 2.]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([2., 0.]),
ng_dissim(centroids, X[1], X=X, membship=membship))
# Unit test for initialization with NaN(i.e., same as matching_dissim)
membship = np.array([[0, 0], [0, 0]])
mdiss_00 = matching_dissim(np.array([X[0]]), np.array([X[0]]))[0]
mdiss_01 = matching_dissim(np.array([X[0]]), np.array([X[1]]))[0]
mdiss_11 = matching_dissim(np.array([X[1]]), np.array([X[1]]))[0]
assert_array_equal(np.array([mdiss_00, mdiss_01]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([mdiss_01, mdiss_11]),
ng_dissim(centroids, X[1], X=X, membship=membship))
X = np.array([['a', 'b', 'c', 'd'], ['a', 'b', 'e', 'd'],
['d', 'c', 'b', 'a']])
centroids = np.array([['a', 'b', 'c', 'd'], ['d', 'c', 'b', 'a']])
membship = np.array([[1, 1, 0], [0, 0, 1]])
assert_array_equal(np.array([0.5, 4.]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([1., 4.]),
ng_dissim(centroids, X[1], X=X, membship=membship))
assert_array_equal(np.array([4., 0.]),
ng_dissim(centroids, X[2], X=X, membship=membship))
# Unit test for initialization (i.e., same as matching_dissim)
membship = np.array([[0, 0, 0], [0, 0, 0]])
mdiss_00 = matching_dissim(np.array([X[0]]), np.array([X[0]]))[0]
mdiss_01 = matching_dissim(np.array([X[0]]), np.array([X[1]]))[0]
mdiss_11 = matching_dissim(np.array([X[1]]), np.array([X[1]]))[0]
mdiss_02 = matching_dissim(np.array([X[0]]), np.array([X[2]]))[0]
mdiss_12 = matching_dissim(np.array([X[0]]), np.array([X[2]]))[0]
mdiss_22 = matching_dissim(np.array([X[2]]), np.array([X[2]]))[0]
assert_array_equal(np.array([mdiss_00, mdiss_02]),
ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([mdiss_01, mdiss_12]),
ng_dissim(centroids, X[1], X=X, membship=membship))
assert_array_equal(np.array([mdiss_12, mdiss_22]),
ng_dissim(centroids, X[2], X=X, membship=membship))
def test_ng_dissim(self):
X = np.array([[0, 1, 2, 0, 1, 2], [0, 1, 2, 0, 1, 1]])
centroids = X
membship = np.array([[1, 0], [0, 1]])
assert_array_equal(np.array([0., 1.]), ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([1., 0.]), ng_dissim(centroids, X[1], X=X, membship=membship))
# Unit test for initialization (i.e., same as matching_dissim)
membship = np.array([[0, 0], [0, 0]])
mdiss_00 = matching_dissim(np.array([X[0]]), np.array([X[0]]))[0]
mdiss_01 = matching_dissim(np.array([X[0]]), np.array([X[1]]))[0]
mdiss_11 = matching_dissim(np.array([X[1]]), np.array([X[1]]))[0]
assert_array_equal(np.array([mdiss_00, mdiss_01]), ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([mdiss_01, mdiss_11]), ng_dissim(centroids, X[1], X=X, membship=membship))
# Unit test for NaN
X = np.array([[np.NaN, 1, 2, 0, 1, 2], [0, 1, 2, 0, 1, 1]])
centroids = X
membship = np.array([[1, 0], [0, 1]])
assert_array_equal(np.array([1., 2.]), ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([2., 0.]), ng_dissim(centroids, X[1], X=X, membship=membship))
# Unit test for initialization with NaN(i.e., same as matching_dissim)
membship = np.array([[0, 0], [0, 0]])
mdiss_00 = matching_dissim(np.array([X[0]]), np.array([X[0]]))[0]
mdiss_01 = matching_dissim(np.array([X[0]]), np.array([X[1]]))[0]
mdiss_11 = matching_dissim(np.array([X[1]]), np.array([X[1]]))[0]
assert_array_equal(np.array([mdiss_00, mdiss_01]), ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([mdiss_01, mdiss_11]), ng_dissim(centroids, X[1], X=X, membship=membship))
X = np.array([['a', 'b', 'c', 'd'], ['a', 'b', 'e', 'd'], ['d', 'c', 'b', 'a']])
centroids = np.array([['a', 'b', 'c', 'd'], ['d', 'c', 'b', 'a']])
membship = np.array([[1, 1, 0], [0, 0, 1]])
assert_array_equal(np.array([0.5, 4.]), ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([1., 4.]), ng_dissim(centroids, X[1], X=X, membship=membship))
assert_array_equal(np.array([4., 0.]), ng_dissim(centroids, X[2], X=X, membship=membship))
# Unit test for initialization (i.e., same as matching_dissim)
membship = np.array([[0, 0, 0], [0, 0, 0]])
mdiss_00 = matching_dissim(np.array([X[0]]), np.array([X[0]]))[0]
mdiss_01 = matching_dissim(np.array([X[0]]), np.array([X[1]]))[0]
mdiss_11 = matching_dissim(np.array([X[1]]), np.array([X[1]]))[0]
mdiss_02 = matching_dissim(np.array([X[0]]), np.array([X[2]]))[0]
mdiss_12 = matching_dissim(np.array([X[0]]), np.array([X[2]]))[0]
mdiss_22 = matching_dissim(np.array([X[2]]), np.array([X[2]]))[0]
assert_array_equal(np.array([mdiss_00, mdiss_02]), ng_dissim(centroids, X[0], X=X, membship=membship))
assert_array_equal(np.array([mdiss_01, mdiss_12]), ng_dissim(centroids, X[1], X=X, membship=membship))
assert_array_equal(np.array([mdiss_12, mdiss_22]), ng_dissim(centroids, X[2], X=X, membship=membship))