def __process_by_ccore(self):
"""!
@brief Performs cluster analysis using CCORE (C/C++ part of pyclustering library).
"""
cure_data_pointer = wrapper.cure_algorithm(
self.__pointer_data, self.__number_cluster,
self.__number_represent_points, self.__compression)
self.__clusters = wrapper.cure_get_clusters(cure_data_pointer)
self.__representors = wrapper.cure_get_representors(cure_data_pointer)
self.__means = wrapper.cure_get_means(cure_data_pointer)
wrapper.cure_data_destroy(cure_data_pointer)
def process(self):
"""!
@brief Performs cluster analysis in line with rules of CURE algorithm.
@remark Results of clustering can be obtained using corresponding get methods.
@see get_clusters()
"""
if (self.__ccore is True):
cure_data_pointer = wrapper.cure_algorithm(
self.__pointer_data, self.__number_cluster,
self.__number_represent_points, self.__compression)
self.__clusters = wrapper.cure_get_clusters(cure_data_pointer)
self.__representors = wrapper.cure_get_representors(
cure_data_pointer)
self.__means = wrapper.cure_get_means(cure_data_pointer)
wrapper.cure_data_destroy(cure_data_pointer)
else:
while (len(self.__queue) > self.__number_cluster):
cluster1 = self.__queue[0]
# cluster that has nearest neighbor.
cluster2 = cluster1.closest
# closest cluster.
#print("Merge decision: \n\t", cluster1, "\n\t", cluster2);
self.__queue.remove(cluster1)
self.__queue.remove(cluster2)
self.__delete_represented_points(cluster1)
self.__delete_represented_points(cluster2)
merged_cluster = self.__merge_clusters(cluster1, cluster2)
self.__insert_represented_points(merged_cluster)
# Pointers to clusters that should be relocated is stored here.
cluster_relocation_requests = []
# Check for the last cluster
if (len(self.__queue) > 0):
merged_cluster.closest = self.__queue[0]
# arbitrary cluster from queue
merged_cluster.distance = self.__cluster_distance(
merged_cluster, merged_cluster.closest)
for item in self.__queue:
distance = self.__cluster_distance(
merged_cluster, item)
# Check if distance between new cluster and current is the best than now.
if (distance < merged_cluster.distance):
merged_cluster.closest = item
merged_cluster.distance = distance
# Check if current cluster has removed neighbor.
if ((item.closest is cluster1)
or (item.closest is cluster2)):
# If previous distance was less then distance to new cluster then nearest cluster should be found in the tree.
#print("Update: ", item);
if (item.distance < distance):
(item.closest,
item.distance) = self.__closest_cluster(
item, distance)
# TODO: investigation of root cause is required.
# Itself and merged cluster should be always in list of neighbors in line with specified radius.
# But merged cluster may not be in list due to error calculation, therefore it should be added
# manually.
if (item.closest is None):
item.closest = merged_cluster
item.distance = distance
# Otherwise new cluster is nearest.
else:
item.closest = merged_cluster
item.distance = distance
cluster_relocation_requests.append(item)
elif (item.distance > distance):
item.closest = merged_cluster
item.distance = distance
cluster_relocation_requests.append(item)
# New cluster and updated clusters should relocated in queue
self.__insert_cluster(merged_cluster)
for item in cluster_relocation_requests:
self.__relocate_cluster(item)
# Change cluster representation
self.__clusters = [
cure_cluster_unit.indexes for cure_cluster_unit in self.__queue
]
self.__representors = [
cure_cluster_unit.rep for cure_cluster_unit in self.__queue
]
self.__means = [
cure_cluster_unit.mean for cure_cluster_unit in self.__queue
]
