def agglomerative_l_method(X, method='ward'):
# library: fastcluster
merge_hist = linkage(X, method=method, metric='euclidean', preserve_input=True)
# reorder to be x [2->N]
num_groups = [i for i in range(2, len(X) + 1)]
merge_dist = list(reversed([each[2] for each in merge_hist]))
cluster_count = refined_l_method(num_groups, merge_dist)
# print('refined_l_method time:', end_time - start_time)
# print('cluster_count:', cluster_count)
# make clusters by merging them according to merge_hist
disjoint = DisjointSet(len(X))
for a, b, _, _ in islice(merge_hist, 0, len(X) - cluster_count):
a, b = int(a), int(b)
disjoint.join(a, b)
# get cluster name for each instance
belong_to = [disjoint.parent(i) for i in range(len(X))]
# print('belong_to:', belong_to)
# counter = Counter(belong_to)
# print('belong_to:', counter)
# rename the cluster name to be 0 -> cluster_count - 1
cluster_map = {}
cluster_name = 0
belong_to_renamed = []
for each in belong_to:
if not each in cluster_map:
cluster_map[each] = cluster_name
cluster_name += 1
belong_to_renamed.append(cluster_map[each])
# print('belong_to_renamed:', belong_to_renamed)
centroids = get_centroids(X, belong_to_renamed)
# print('centroids:', centroids)
return Result(belong_to_renamed, centroids)
def fit(self, X, max_merge_dist):
self.X = X
self.max_merge_dist = max_merge_dist
merge_hist = linkage(X, method='ward', metric='euclidean', preserve_input=True)
disjoint = DisjointSet(len(X))
# _, _, merge_dists, _ = list(zip(*merge_hist))
# print('merge_dists:', merge_dists)
for a, b, merge_dist, _ in merge_hist:
if merge_dist > max_merge_dist:
break
a, b = int(a), int(b)
disjoint.join(a, b)
belong_to = [disjoint.parent(i) for i in range(len(X))]
# rename the cluster name to be 0 -> cluster_count - 1
cluster_map = {}
cluster_name = 0
belong_to_renamed = []
for each in belong_to:
if not each in cluster_map:
cluster_map[each] = cluster_name
cluster_name += 1
belong_to_renamed.append(cluster_map[each])
# print('belong_to_renamed:', belong_to_renamed)
centroids, cluster_member_cnt = self.get_centroids(X, belong_to_renamed)
self.cluster_centers_ = centroids
# print('centroids:', centroids)
# print('cluster_member_cnt:', cluster_member_cnt)
return centroids, cluster_member_cnt
