def test_fit_predict_default(data):
v_data = data['fit_data'][:2]
spectral = MultiviewSpectralClustering(2, random_state=RANDOM_STATE)
predictions = spectral.fit_predict(v_data)
n_clusts = data['n_clusters']
assert (predictions.shape[0] == data['n_fit'])
for clust in predictions:
assert (clust >= 0 and clust < n_clusts)
def test_fit_predict_info_view(data):
v_data = data['fit_data']
info_view = np.random.randint(len(v_data))
n_clusts = data['n_clusters']
spectral = MultiviewSpectralClustering(n_clusts,
random_state=RANDOM_STATE,
info_view=info_view)
predictions = spectral.fit_predict(v_data)
assert (predictions.shape[0] == data['n_fit'])
for clust in predictions:
assert (clust >= 0 and clust < n_clusts)
def test_fit_predict_max_iter(data):
v_data = data['fit_data']
max_iter = 5
n_clusts = data['n_clusters']
spectral = MultiviewSpectralClustering(n_clusts,
random_state=RANDOM_STATE,
max_iter=max_iter)
predictions = spectral.fit_predict(v_data)
assert (predictions.shape[0] == data['n_fit'])
for clust in predictions:
assert (clust >= 0 and clust < n_clusts)
def test_not_valid_affinity(small_data):
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(affinity='What')
spectral.fit_predict(small_data)
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(affinity=None)
spectral.fit_predict(small_data)
def test_n_clusters_not_positive_int(small_data):
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(n_clusters=-1)
spectral.fit_predict(small_data)
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(n_clusters=0)
spectral.fit_predict(small_data)
def test_info_view_not_valid(small_data):
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(n_clusters=2, info_view=-1)
spectral.fit_predict(small_data)
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(n_clusters=2, info_view=6)
spectral.fit_predict(small_data)
def test_gamma_not_positive_float(small_data):
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(gamma=-1.5)
spectral.fit_predict(small_data)
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(gamma=0)
spectral.fit_predict(small_data)
def test_n_neighbors_not_positive_int(small_data):
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(affinity='nearest_neighbors',
n_neighbors=-1)
spectral.fit_predict(small_data)
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(affinity='nearest_neighbors',
n_neighbors=0)
spectral.fit_predict(small_data)
def test_affinity_neighbors(data):
v1_data = data['fit_data'][0]
n_neighbors = 10
neighbors = NearestNeighbors(n_neighbors=n_neighbors)
neighbors.fit(v1_data)
true_kernel = neighbors.kneighbors_graph(v1_data).toarray()
spectral = MultiviewSpectralClustering(random_state=RANDOM_STATE,
affinity='nearest_neighbors',
n_neighbors=10)
n_kernel = spectral._affinity_mat(v1_data)
assert (n_kernel.shape[0] == data['n_fit'])
assert (n_kernel.shape[1] == data['n_fit'])
for ind1 in range(n_kernel.shape[0]):
for ind2 in range(n_kernel.shape[1]):
assert np.abs(true_kernel[ind1][ind2] -
n_kernel[ind1][ind2]) < 0.000001
def test_affinity_mat_poly(data):
v1_data = data['fit_data'][0]
distances = cdist(v1_data, v1_data)
gamma = 1 / (2 * np.median(distances)**2)
true_kernel = polynomial_kernel(v1_data, gamma=gamma)
spectral = MultiviewSpectralClustering(random_state=RANDOM_STATE,
affinity='poly')
p_kernel = spectral._affinity_mat(v1_data)
assert (p_kernel.shape[0] == data['n_fit'])
assert (p_kernel.shape[1] == data['n_fit'])
for ind1 in range(p_kernel.shape[0]):
for ind2 in range(p_kernel.shape[1]):
assert np.abs(true_kernel[ind1][ind2] -
p_kernel[ind1][ind2]) < 0.000001
def test_affinity_mat_rbf2(data):
v1_data = data['fit_data'][0]
gamma = 1
spectral = MultiviewSpectralClustering(random_state=RANDOM_STATE,
gamma=gamma)
distances = cdist(v1_data, v1_data)
gamma = 1 / (2 * np.median(distances)**2)
true_kernel = rbf_kernel(v1_data, gamma=1)
g_kernel = spectral._affinity_mat(v1_data)
assert (g_kernel.shape[0] == data['n_fit'])
assert (g_kernel.shape[1] == data['n_fit'])
for ind1 in range(g_kernel.shape[0]):
for ind2 in range(g_kernel.shape[1]):
assert np.abs(true_kernel[ind1][ind2] -
g_kernel[ind1][ind2]) < 0.000001
def perform_clustering(seed, m_data, labels, n_clusters, kernel='rbf'):
# Single-view spectral clustering
# Cluster each view separately
s_spectral = SpectralClustering(n_clusters=n_clusters,
random_state=RANDOM_SEED,
affinity=kernel,
n_init=100)
s_clusters_v1 = s_spectral.fit_predict(m_data[0])
s_clusters_v2 = s_spectral.fit_predict(m_data[1])
# Concatenate the multiple views into a single view
s_data = np.hstack(m_data)
s_clusters = s_spectral.fit_predict(s_data)
# Compute nmi between true class labels and single-view cluster labels
s_nmi_v1 = nmi_score(labels, s_clusters_v1)
s_nmi_v2 = nmi_score(labels, s_clusters_v2)
s_nmi = nmi_score(labels, s_clusters)
print('Single-view View 1 NMI Score: {0:.3f}\n'.format(s_nmi_v1))
print('Single-view View 2 NMI Score: {0:.3f}\n'.format(s_nmi_v2))
print('Single-view Concatenated NMI Score: {0:.3f}\n'.format(s_nmi))
# Multi-view spectral clustering
# Use the MultiviewSpectralClustering instance to cluster the data
m_spectral = MultiviewSpectralClustering(n_clusters=n_clusters,
random_state=RANDOM_SEED,
affinity=kernel,
n_init=100)
m_clusters = m_spectral.fit_predict(m_data)
# Compute nmi between true class labels and multi-view cluster labels
m_nmi = nmi_score(labels, m_clusters)
print('Multi-view Concatenated NMI Score: {0:.3f}\n'.format(m_nmi))
return m_clusters
def data():
num_fit_samples = 200
n_feats1 = 20
n_feats2 = 18
n_feats3 = 30
n_clusters = 2
np.random.seed(RANDOM_STATE)
fit_data = []
fit_data.append(np.random.rand(num_fit_samples, n_feats1))
fit_data.append(np.random.rand(num_fit_samples, n_feats2))
fit_data.append(np.random.rand(num_fit_samples, n_feats3))
spectral = MultiviewSpectralClustering(n_clusters,
random_state=RANDOM_STATE)
return {
'n_fit': num_fit_samples,
'n_feats1': n_feats1,
'n_feats2': n_feats2,
'n_feats3': n_feats3,
'n_clusters': n_clusters,
'spectral': spectral,
'fit_data': fit_data
}
# Compute nmi between true class labels and single-view cluster labels
s_nmi_v1 = nmi_score(labels, s_clusters_v1)
s_nmi_v2 = nmi_score(labels, s_clusters_v2)
s_nmi = nmi_score(labels, s_clusters)
print('Single-view View 1 NMI Score: {0:.3f}\n'.format(s_nmi_v1))
print('Single-view View 2 NMI Score: {0:.3f}\n'.format(s_nmi_v2))
print('Single-view Concatenated NMI Score: {0:.3f}\n'.format(s_nmi))
###############################################################################
# Multiview spectral clustering
# ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
#
# Use the MultiviewSpectralClustering instance to cluster the data
m_spectral = MultiviewSpectralClustering(n_clusters=n_class,
affinity='nearest_neighbors',
max_iter=12,
random_state=RANDOM_SEED,
n_init=10)
m_clusters = m_spectral.fit_predict(m_data)
# Compute nmi between true class labels and multi-view cluster labels
m_nmi = nmi_score(labels, m_clusters)
print('Multi-view NMI Score: {0:.3f}\n'.format(m_nmi))
###############################################################################
# Plots of clusters produced by multi-view spectral clustering and the true
# clusters
#
# We will display the clustering results of the Multi-view spectral clustering
# algorithm below, along with the true class labels.
sca_kwargs = {'alpha': 0.7, 's': 10}
f, axes = plt.subplots(1, 2, figsize=(8, 4))
axes[0].scatter(Xs[:, 0], Xs[:, 1], c=y_true, **sca_kwargs)
axes[0].set_title('True labels', fontsize=14)
axes[1].scatter(Xs[:, 0], Xs[:, 1], c=y_predicted, **sca_kwargs)
axes[1].set_title(title, fontsize=14)
axes[1].annotate(f'Homogeneity\nscore = {score:.2f}',
xy=(0.95, 0.85),
xycoords='axes fraction',
fontsize=13,
ha='right')
axes[0].set_ylabel(f'{method} Component 2')
plt.setp(axes, xticks=[], yticks=[], xlabel=f'{method} Component 1')
plt.tight_layout()
plt.show()
# Cluster concatenated data
sv_clust = SpectralClustering(n_clusters=4, affinity='nearest_neighbors')
sv_labels = sv_clust.fit_predict(np.hstack(Xs))
plot_clusters(Xs_pca, y, sv_labels, 'Concatenated clustering labels', 'PCA')
# Cluster multiview data
mv_clust = MultiviewSpectralClustering(n_clusters=4,
affinity='nearest_neighbors')
mv_labels = mv_clust.fit_predict(Xs)
plot_clusters(Xs_mvmds, y, mv_labels, 'Multiview clustering labels', 'MVMDS')
def test_samples_not_2D_2(small_data):
with pytest.raises(ValueError):
view1 = np.random.random((10, ))
view2 = np.random.random((10, ))
spectral = MultiviewSpectralClustering(random_state=RANDOM_STATE)
spectral.fit_predict([view1, view2])
def test_n_views_too_small2(small_data):
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(random_state=RANDOM_STATE)
spectral.fit_predict([])
def test_random_state_not_convertible(small_data):
with pytest.raises(ValueError):
spectral = MultiviewSpectralClustering(n_clusters=5, random_state='ab')
spectral.fit_predict(small_data)