def cluster(self, name='kmeans', min_clusters=2, max_clusters=None, initial='k-means++'):
if min_clusters < 2:
print 'Must have at least 2 clusters'
min_clusters = 2
if name != 'kmeans' and name != 'kmedoids':
print 'Invalid clustering algorithm name. Defaulting to k-means'
name='kmeans'
if max_clusters == None:
max_clusters = len(self.data)/2.0
max = -2
num = 0
labely = []
r = max_clusters - min_clusters+1
if name == 'kmedoids':
for i in range(r):
num_clusters = i + min_clusters
cl = kmedoids(self.points, num_clusters)
self.labels = [0] * len(self.data)
cluster = -1
for key in cl[2]:
cluster += 1
for j in cl[2][key]:
self.labels[j] = cluster
sil = metrics.silhouette_score(numpy.array(self.points), numpy.array(self.labels))
if sil > max:
max = sil
num = num_clusters
labely = self.labels
elif name == 'kmeans':
for i in range(r):
num_clusters = i + min_clusters
cl = KMeans(num_clusters, random_state=8, init=initial)
cl.fit(self.points)
self.labels = cl.labels_
sil = metrics.silhouette_score(numpy.array(self.points), self.labels)
if sil > max:
max = sil
num = num_clusters
labely = self.labels
self.sil = max
self.clusters = num
self.labels = labely
def cluster(self, name='kmeans', min_clusters=2, max_clusters=None):
logging.debug("min_clusters = %s, max_clusters = %s, len(self.points) = %s" %
(min_clusters, max_clusters, len(self.points)))
if min_clusters < 2:
logging.debug("min_clusters < 2, setting min_clusters = 2")
min_clusters = 2
if min_clusters > len(self.points):
sys.stderr.write('Maximum number of clusters is the number of data points.\n')
min_clusters = len(self.points)-1
if max_clusters == None:
logging.debug("max_clusters is None, setting max_clusters = %d" % (len(self.points) - 1))
max_clusters = len(self.points)-1
if max_clusters < 2:
sys.stderr.write('Must have at least 2 clusters\n')
max_clusters = 2
if max_clusters >= len(self.points):
logging.debug("max_clusters >= len(self.points), setting max_clusters = %d" % (len(self.points) - 1))
max_clusters = len(self.points)-1
if max_clusters < min_clusters:
raise ValueError('Please provide a valid range of cluster sizes')
if name != 'kmeans' and name != 'kmedoids':
logging.debug('Invalid clustering algorithm name. Defaulting to k-means')
name='kmeans'
if not self.data:
self.sil = None
self.clusters = 0
self.labels = []
return []
max = -2
num = 0
labely = []
r = max_clusters - min_clusters+1
if name == 'kmedoids':
for i in range(r):
num_clusters = i + min_clusters
logging.debug('testing %s clusters' % str(num_clusters))
cl = kmedoids(self.points, num_clusters)
self.labels = [0] * len(self.data)
cluster = -1
for key in cl[2]:
cluster += 1
for j in cl[2][key]:
self.labels[j] = cluster
sil = metrics.silhouette_score(numpy.array(self.points), numpy.array(self.labels))
if sil > max:
max = sil
num = num_clusters
labely = self.labels
elif name == 'kmeans':
import warnings
for i in range(r):
num_clusters = i + min_clusters
cl = KMeans(num_clusters, random_state=8)
cl.fit(self.points)
self.labels = cl.labels_
warnings.filterwarnings("ignore")
sil = metrics.silhouette_score(numpy.array(self.points), self.labels)
if sil > max:
max = sil
num = num_clusters
labely = self.labels
self.sil = max
self.clusters = num
self.labels = labely
self.labels = list(self.labels)
return self.labels
def cluster(self, name='kmeans', min_clusters=2, max_clusters=None):
logging.debug(
"min_clusters = %s, max_clusters = %s, len(self.points) = %s" %
(min_clusters, max_clusters, len(self.points)))
if min_clusters < 2:
logging.debug("min_clusters < 2, setting min_clusters = 2")
min_clusters = 2
if min_clusters > len(self.points):
sys.stderr.write(
'Minimum number of clusters %d is greater than the number of data points %d.\n'
% (min_clusters, len(self.points)))
min_clusters = len(self.points) - 1
if max_clusters == None:
logging.debug("max_clusters is None, setting max_clusters = %d" %
(len(self.points) - 1))
max_clusters = len(self.points) - 1
if max_clusters < 2:
sys.stderr.write('Must have at least 2 clusters\n')
max_clusters = 2
if max_clusters >= len(self.points):
logging.debug(
"max_clusters >= len(self.points), setting max_clusters = %d" %
(len(self.points) - 1))
max_clusters = len(self.points) - 1
if max_clusters < min_clusters:
raise ValueError('Please provide a valid range of cluster sizes')
if name != 'kmeans' and name != 'kmedoids':
logging.debug(
'Invalid clustering algorithm name. Defaulting to k-means')
name = 'kmeans'
if not self.data:
self.sil = None
self.clusters = 0
self.labels = []
return []
max = -2
num = 0
labely = []
r = max_clusters - min_clusters + 1
if name == 'kmedoids':
for i in range(r):
num_clusters = i + min_clusters
logging.debug('testing %s clusters' % str(num_clusters))
cl = kmedoids(self.points, num_clusters)
self.labels = [0] * len(self.data)
cluster = -1
for key in cl[2]:
cluster += 1
for j in cl[2][key]:
self.labels[j] = cluster
sil = metrics.silhouette_score(numpy.array(self.points),
numpy.array(self.labels))
if sil > max:
max = sil
num = num_clusters
labely = self.labels
elif name == 'kmeans':
import warnings
for i in range(r):
num_clusters = i + min_clusters
if num_clusters == 0:
continue
cl = KMeans(num_clusters, random_state=8)
cl.fit(self.points)
self.labels = cl.labels_
warnings.filterwarnings("ignore")
sil = metrics.silhouette_score(numpy.array(self.points),
self.labels)
if sil > max:
max = sil
num = num_clusters
labely = self.labels
self.sil = max
self.clusters = num
self.labels = labely
self.labels = list(self.labels)
return self.labels