def test_KMeansConstrained_parity_digits():
iris = datasets.load_iris()
X = iris.data
k = 8
random_state = 1
size_min, size_max = None, None # No restrictions and so should produce same result
clf_constrained = KMeansConstrained(
n_clusters=k,
size_min=size_min,
size_max=size_max,
random_state=random_state
)
y_constrained = clf_constrained.fit_predict(X)
clf_kmeans = KMeans(
n_clusters=k,
random_state=random_state
)
y_kmeans = clf_kmeans.fit_predict(X)
assert_array_equal(y_constrained, y_kmeans)
assert_almost_equal(clf_constrained.cluster_centers_, clf_kmeans.cluster_centers_)
assert_almost_equal(clf_constrained.inertia_, clf_kmeans.inertia_)
def test_KMeansConstrained_predict_method():
X = np.array([
[0, 0],
[0, 0],
[0, 0],
[1, 1],
])
k = 2
size_max = 2
clf = KMeansConstrained(n_clusters=k, size_max=size_max)
clf.fit(X)
y_constrained = clf.predict(X) # Expected np.array([0, 0, 1, 1])
y_normal = super(KMeansConstrained,
clf).predict(X) # Expected np.array([0, 0, 0, 1])
cluster_size_constrained = pd.Series(y_constrained).value_counts()
assert (cluster_size_constrained > size_max).any() == False
assert len(cluster_size_constrained) == k
cluster_size_normal = pd.Series(y_normal).value_counts()
assert (cluster_size_normal > size_max).any() == True
assert len(cluster_size_normal) == k
def test_KMeansConstrained():
X = np.array([
[0, 0],
[1, 2],
[1, 4],
[1, 0],
[4, 2],
[4, 4],
[4, 0],
[3, 0],
[4, 4]
])
k = 3
size_min, size_max = 3, 7
clf = KMeansConstrained(
n_clusters=k,
size_min=size_min,
size_max=size_max
)
clf.fit(X)
y = clf.fit_predict(X)
# Labels
cluster_size = pd.Series(y).value_counts()
assert (cluster_size > size_max).sum() == 0
assert (cluster_size < size_min).sum() == 0
def test_KMeansConstrained_performance():
n_cluster = 10
n_X = 1000
d = 3
seed = 1
np.random.seed(seed=seed)
X = np.random.rand(n_X, d)
clf = KMeansConstrained(n_cluster, size_min=None, size_max=None,
init='k-means++', n_init=10, max_iter=300, tol=1e-4,
verbose=False, random_state=seed, copy_x=True, n_jobs=1)
y = clf.fit_predict(X)
#time = timeit('y = clf.fit_predict(X)', number=1, globals=globals())
def test_KMeansConstrained_n_jobs():
X, _ = datasets.make_blobs(n_samples=100, n_features=5, centers=10, random_state=1)
n_jobs = -1
k = 20
size_min, size_max = 3, 40
clf = KMeansConstrained(
n_clusters=k,
size_min=size_min,
size_max=size_max,
n_jobs=n_jobs
)
clf.fit(X)
y = clf.fit_predict(X)
# Labels
cluster_size = pd.Series(y).value_counts()
assert (cluster_size > size_max).sum() == 0
assert (cluster_size < size_min).sum() == 0
def test_KMeansConstrained_parity_digits():
iris = datasets.load_iris()
X = iris.data
k = 8
random_state = 1
size_min, size_max = None, None # No restrictions and so should produce same result
clf_constrained = KMeansConstrained(size_min=size_min,
size_max=size_max,
n_clusters=k,
random_state=random_state,
init='k-means++',
n_init=10,
max_iter=300,
tol=1e-4)
y_constrained = clf_constrained.fit_predict(X)
# TODO: Testing scikit-learn has be set to v0.19. This is because there is a discrepancy scikit-learn v0.22 https://github.com/scikit-learn/scikit-learn/issues/16623
clf_kmeans = KMeans(n_clusters=k,
random_state=random_state,
init='k-means++',
n_init=10,
max_iter=300,
tol=1e-4)
y_kmeans = clf_kmeans.fit_predict(X)
# Each cluster should have the same number of datapoints assigned to it
constrained_ndp = pd.Series(y_constrained).value_counts().values
kmeans_ndp = pd.Series(y_kmeans).value_counts().values
assert_almost_equal(constrained_ndp, kmeans_ndp)
# Sort the cluster coordinates (otherwise in a random order)
constrained_cluster_centers = sort_coordinates(
clf_constrained.cluster_centers_)
kmean_cluster_centers = sort_coordinates(clf_kmeans.cluster_centers_)
assert_almost_equal(constrained_cluster_centers, kmean_cluster_centers)
def test_spare_not_implemented():
X = np.array([[0, 0], [1, 2], [1, 4], [1, 0], [4, 2], [4, 4], [4, 0],
[3, 0], [4, 4]])
k = 3
size_min, size_max = 3, 7
clf = KMeansConstrained(n_clusters=k, size_min=size_min, size_max=size_max)
X = csc_matrix(X)
with pytest.raises(NotImplementedError):
clf.fit(X)
with pytest.raises(NotImplementedError):
clf.fit_predict(X)
#######
# Parity tests only works with sklearn v0.19.2 but does not run on Python 3.8+
#######
# from sklearn.cluster import KMeans
# from sklearn.cluster.k_means_ import _labels_inertia
# from numpy.testing import assert_array_equal, assert_almost_equal
# from k_means_constrained.sklearn_import.utils.extmath import row_norms
# Test passes on Python 3.7
# def test__labels_constrained_kmeans_parity():
# X = np.array([
# [0, 0],
# [1, 2],
# [1, 4],
# [1, 0],
# [4, 2],
# [4, 4],
# [4, 0],
# [4, 4]
# ]).astype('float')
# centers = np.array([
# [0, 0],
# [4, 4]
# ]).astype('float')
# size_min, size_max = 0, len(X) # No restrictions and so should be the same as K-means
#
# x_squared_norms = row_norms(X, squared=True)
#
# distances_constrained = np.zeros(shape=(X.shape[0],), dtype=X.dtype)
# labels_constrained, inertia_constrained = _labels_constrained(X, centers, size_min, size_max, distances_constrained)
#
# distances_kmeans = np.zeros(shape=(X.shape[0],), dtype=X.dtype)
# labels_kmeans, inertia_kmeans = \
# _labels_inertia(X=X, x_squared_norms=x_squared_norms, centers=centers, precompute_distances=False,
# distances=distances_kmeans)
#
# assert_array_equal(labels_constrained, labels_kmeans)
# assert_almost_equal(distances_constrained, distances_kmeans)
# assert inertia_constrained == inertia_kmeans
# Test passes on Python 3.7
# def test_KMeansConstrained_parity_digits():
# iris = datasets.load_iris()
# X = iris.data
#
# k = 8
# random_state = 1
# size_min, size_max = None, None # No restrictions and so should produce same result
#
# clf_constrained = KMeansConstrained(
# size_min=size_min,
# size_max=size_max,
# n_clusters=k,
# random_state=random_state,
# init='k-means++',
# n_init=10,
# max_iter=300,
# tol=1e-4
# )
# y_constrained = clf_constrained.fit_predict(X)
#
# # TODO: Testing scikit-learn has be set to v0.19. This is because there is a discrepancy scikit-learn v0.22 https://github.com/scikit-learn/scikit-learn/issues/16623
# clf_kmeans = KMeans(
# n_clusters=k,
# random_state=random_state,
# init='k-means++',
# n_init=10,
# max_iter=300,
# tol=1e-4
# )
# y_kmeans = clf_kmeans.fit_predict(X)
#
# # Each cluster should have the same number of datapoints assigned to it
# constrained_ndp = pd.Series(y_constrained).value_counts().values
# kmeans_ndp = pd.Series(y_kmeans).value_counts().values
#
# assert_almost_equal(constrained_ndp, kmeans_ndp)
#
# # Sort the cluster coordinates (otherwise in a random order)
# constrained_cluster_centers = sort_coordinates(clf_constrained.cluster_centers_)
# kmean_cluster_centers = sort_coordinates(clf_kmeans.cluster_centers_)
#
# assert_almost_equal(constrained_cluster_centers, kmean_cluster_centers)
####
# Further tests removed as removed sklearn dependency
####
# from sklearn import datasets
#
# def test_KMeansConstrained_n_jobs():
# X, _ = datasets.make_blobs(n_samples=100, n_features=5, centers=10, random_state=1)
#
# n_jobs = -1
# k = 20
# size_min, size_max = 3, 40
#
# clf = KMeansConstrained(
# n_clusters=k,
# size_min=size_min,
# size_max=size_max,
# n_jobs=n_jobs
# )
#
# y = clf.fit_predict(X)
#
# # Labels
# cluster_size = pd.Series(y).value_counts()
# assert (cluster_size > size_max).sum() == 0
# assert (cluster_size < size_min).sum() == 0