def testClustersAssignments(self):
# Ignore noise
ignore_noise = True
num_categories = self.num_categories - 1
(cluster_assignments,
sdr_slices) = find_cluster_assignments(self.sdrs, self.clusters,
ignore_noise)
self.assertEqual(len(cluster_assignments), num_categories * self.reps)
self.assertEqual(len(sdr_slices), num_categories * self.reps)
# Don't ignore the noise
ignore_noise = False
num_categories = self.num_categories
(cluster_assignments,
sdr_slices) = find_cluster_assignments(self.sdrs, self.clusters,
ignore_noise)
self.assertEqual(len(cluster_assignments), num_categories * self.reps)
self.assertEqual(len(sdr_slices), num_categories * self.reps)
def plot_inter_sequence_distances(output_dir, plot_id, distance_func, sdrs,
cluster_ids, ignore_noise):
cluster_assignments, sdr_slices = find_cluster_assignments(
sdrs, cluster_ids, ignore_noise)
distance_mat = cluster_distance_matrix(sdr_slices, distance_func)
title = 'distance_matrix_%s' % plot_id
outputFile = '%s/%s' % (output_dir, '%s.png' % title)
plot_distance_mat(distance_mat, title, outputFile)
def testClustersAssignments(self):
# Ignore noise
ignore_noise = True
num_categories = self.num_categories - 1
(cluster_assignments,
sdr_slices) = find_cluster_assignments(self.sdrs,
self.clusters,
ignore_noise)
self.assertEqual(len(cluster_assignments), num_categories * self.reps)
self.assertEqual(len(sdr_slices), num_categories * self.reps)
# Don't ignore the noise
ignore_noise = False
num_categories = self.num_categories
(cluster_assignments,
sdr_slices) = find_cluster_assignments(self.sdrs,
self.clusters,
ignore_noise)
self.assertEqual(len(cluster_assignments), num_categories * self.reps)
self.assertEqual(len(sdr_slices), num_categories * self.reps)
def plot_inter_sequence_distances(output_dir,
plot_id,
distance_func,
sdrs,
cluster_ids,
ignore_noise):
cluster_assignments, sdr_slices = find_cluster_assignments(sdrs, cluster_ids,
ignore_noise)
distance_mat = cluster_distance_matrix(sdr_slices, distance_func)
title = 'distance_matrix_%s' % plot_id
outputFile = '%s/%s' % (output_dir, '%s.png' % title)
plot_distance_mat(distance_mat, title, outputFile)
def main():
# OPTICS demo with SDRs
radius = 10.0
neighbors = 2
epsilon_cutoff = 0.3
file_path = os.path.join(
os.getcwd(), 'htm_traces', 'binary_ampl=10.0_mean=0.0_noise=0.0_'
'sp=True_tm=True_tp=False_SDRClassifier.csv')
sdrs, cluster_ids = load_sdrs(file_path)
(cluster_assignments,
sdr_slices) = find_cluster_assignments(sdrs,
cluster_ids,
ignore_noise=True)
sdr_cluster_centroids = get_sdr_cluster_centroids(sdr_slices)
# Project SDRs in 2D for visualization purposes
distance_mat = distance_matrix(sdr_cluster_centroids, euclidian_distance)
sdr_projections = project_clusters_2D(distance_mat, method='mds')
# OPTICS SDR results
sample_sdr = sdr_projections # or: sdr_cluster_centroids
ordering_sdr, clusters_sdr, noise_sdr = analyze(sample_sdr, radius,
neighbors)
# Plot OPTICS results
sample_sdr = sdr_projections
plot_results(sample_sdr, ordering_sdr, clusters_sdr, noise_sdr, 'SDR',
epsilon_cutoff)
# Plot SDR 2D projections
title = '2d_projections'
output_file = '%s' % ('%s.png' % title)
plt = plot_2D_projections(title, output_file, cluster_assignments,
sdr_projections)
plt.show()
# OPTICS demo with 2D vectors
radius = 2.0
neighbors = 2
epsilon_cutoff = 0.5
sample_2d = read_sample(
FCPS_SAMPLES.SAMPLE_LSUN) # or: gaussian_clusters(3)
ordering_2d, clusters_2d, noise_2d = analyze(sample_2d, radius, neighbors)
plot_results(sample_2d, ordering_2d, clusters_2d, noise_2d, '2D',
epsilon_cutoff)
# Plot input data and clustering structure
ordering_visualizer.show_ordering_diagram(ordering_sdr)
ordering_visualizer.show_ordering_diagram(ordering_2d)
def plot_inter_sequence_distances(output_dir, plot_id, distance_func, sdrs,
cluster_ids, ignore_noise):
cluster_assignments, sdr_slices = find_cluster_assignments(
sdrs, cluster_ids, ignore_noise)
distance_mat = cluster_distance_matrix(sdr_slices, distance_func)
title = 'distance_matrix_%s' % plot_id
output_file = '%s/%s' % (output_dir, '%s.png' % title)
plot_distance_mat(distance_mat, title, output_file)
projections = project_clusters_2D(distance_mat, method='mds')
title = '2d_projections_%s' % plot_id
output_file = '%s/%s' % (output_dir, '%s.png' % title)
plot_2D_projections(title, output_file, cluster_assignments, projections)
def main():
# OPTICS demo with SDRs
radius = 10.0
neighbors = 2
epsilon_cutoff = 0.3
file_path = os.path.join(os.getcwd(), 'htm_traces',
'binary_ampl=10.0_mean=0.0_noise=0.0_'
'sp=True_tm=True_tp=False_SDRClassifier.csv')
sdrs, cluster_ids = load_sdrs(file_path)
(cluster_assignments,
sdr_slices) = find_cluster_assignments(sdrs, cluster_ids, ignore_noise=True)
sdr_cluster_centroids = get_sdr_cluster_centroids(sdr_slices)
# Project SDRs in 2D for visualization purposes
distance_mat = distance_matrix(sdr_cluster_centroids, euclidian_distance)
sdr_projections = project_clusters_2D(distance_mat, method='mds')
# OPTICS SDR results
sample_sdr = sdr_projections # or: sdr_cluster_centroids
ordering_sdr, clusters_sdr, noise_sdr = analyze(sample_sdr,
radius,
neighbors)
# Plot OPTICS results
sample_sdr = sdr_projections
plot_results(sample_sdr, ordering_sdr, clusters_sdr, noise_sdr,
'SDR', epsilon_cutoff)
# Plot SDR 2D projections
title = '2d_projections'
output_file = '%s' % ('%s.png' % title)
plt = plot_2D_projections(title, output_file, cluster_assignments,
sdr_projections)
plt.show()
# OPTICS demo with 2D vectors
radius = 2.0
neighbors = 2
epsilon_cutoff = 0.5
sample_2d = read_sample(FCPS_SAMPLES.SAMPLE_LSUN) # or: gaussian_clusters(3)
ordering_2d, clusters_2d, noise_2d = analyze(sample_2d, radius, neighbors)
plot_results(sample_2d, ordering_2d, clusters_2d, noise_2d, '2D',
epsilon_cutoff)
# Plot input data and clustering structure
ordering_visualizer.show_ordering_diagram(ordering_sdr)
ordering_visualizer.show_ordering_diagram(ordering_2d)
def plot_inter_sequence_distances(output_dir,
plot_id,
distance_func,
sdrs,
cluster_ids,
ignore_noise):
cluster_assignments, sdr_slices = find_cluster_assignments(sdrs, cluster_ids,
ignore_noise)
distance_mat = cluster_distance_matrix(sdr_slices, distance_func)
title = 'distance_matrix_%s' % plot_id
output_file = '%s/%s' % (output_dir, '%s.png' % title)
plot_distance_mat(distance_mat, title, output_file)
projections = project_clusters_2D(distance_mat, method='mds')
title = '2d_projections_%s' % plot_id
output_file = '%s/%s' % (output_dir, '%s.png' % title)
plot_2D_projections(title, output_file, cluster_assignments, projections)
