def cluster_rotate(
self,
eigenvectors,
max_groups,
min_groups=2,
tolerance=0.0025,
):
groups = range(min_groups, max_groups + 1)
vector_length = eigenvectors.shape[0]
current_vector = eigenvectors[:, :groups[0]]
n = max_groups - min_groups + 1
quality_scores = [None] * n
clusters = [None] * n
rotated_vectors = [None] * n
for g in range(n):
if g > 0:
current_vector = np.concatenate((rotated_vectors[g - 1],
eigenvectors[:, groups[g] - 1:groups[g]]), axis=1)
(clusters[g], quality_scores[g], rotated_vectors[g]) = \
evrot.main(current_vector)
return (groups, clusters, quality_scores, rotated_vectors)
def cluster_rotate(
self,
eigenvectors,
max_groups,
min_groups=2,
tolerance=0.0025,
):
groups = range(min_groups, max_groups + 1)
vector_length = eigenvectors.shape[0]
current_vector = eigenvectors[:, :groups[0]]
n = max_groups - min_groups + 1
quality_scores = [None] * n
clusters = [None] * n
rotated_vectors = [None] * n
for g in range(n):
if g > 0:
current_vector = np.concatenate(
(rotated_vectors[g - 1],
eigenvectors[:, groups[g] - 1:groups[g]]),
axis=1)
(clusters[g], quality_scores[g], rotated_vectors[g]) = \
evrot.main(current_vector)
# Find the highest index of quality scores where the
# score is within 0.0025 of the maximum:
# this is our chosen number of groups
max_score = max(quality_scores)
index = quality_scores.index(max_score)
start = index + 1
for (i, score) in enumerate(quality_scores[index + 1:], start=start):
if abs(score - max_score) < tolerance:
index = i
return (groups[index], clusters[index], quality_scores,
rotated_vectors[index])
def cluster_rotate(
self,
eigenvectors,
max_groups,
min_groups=2,
):
groups = range(min_groups, max_groups + 1)
vector_length = eigenvectors.shape[0]
current_vector = eigenvectors[:, :groups[0]]
n = max_groups - min_groups + 1
quality_scores = [None] * n
clusters = [None] * n
rotated_vectors = [None] * n
for g in range(n):
if g > 0:
current_vector = np.concatenate((rotated_vectors[g
- 1], eigenvectors[:, groups[g] - 1:
groups[g]]), axis=1)
(clusters[g], quality_scores[g], rotated_vectors[g]) = \
evrot.main(current_vector)
# Find the highest index of quality scores where the
# score is within 0.0025 of the maximum:
# this is our chosen number of groups
max_score = max(quality_scores)
index = quality_scores.index(max_score)
start = index + 1
for (i, score) in enumerate(quality_scores[index + 1:],
start=start):
if abs(score - max_score) < 0.0025:
index = i
return (groups[index], clusters[index], quality_scores,
rotated_vectors[index])