Python实现简单层次聚类算法以及可视化

import numpy as np
import data_helper
np.random.seed(1)
def get_raw_data(n):
 _data=np.random.rand(n,2)
 #生成数据的格式是n个（x,y）
 _groups={idx:[[x,y]] for idx,(x,y) in enumerate(_data)}
 return _groups
def cal_distance(cluster1,cluster2):
 #采用最小距离作为聚类标准
 _min_distance=10000
 for x1,y1 in cluster1:
  for x2,y2 in cluster2:
   _distance=(x1-x2)**2+(y1-y2)**2
   if _distance<_min_distance:
    _min_distance=_distance
 return _distance
groups=get_raw_data(10)
count=0
while len(groups)!=1:#判断是不是所有的数据是不是归为了同一类
 min_distance=10000
 len_groups=len(groups)
 
 for i in groups.keys():
  for j in groups.keys():
   if i>=j:
    continue
   distance=cal_distance(groups[i],groups[j])
   if distance<min_distance:
    min_distance=distance
    min_i=i
    min_j=j#这里的j>i
 groups[min_i].extend(groups.pop(min_j))
 data_helper.draw_data(groups)
 #一共n个簇，共迭代n-1次