您的位置:首页 > 脚本大全 > > 正文

python 聚类找出同一类别的数据(Python实现简单层次聚类算法以及可视化)

更多 时间:2021-11-05 14:02:59 类别:脚本大全 浏览量:731

python 聚类找出同一类别的数据

Python实现简单层次聚类算法以及可视化

本文实例为大家分享了Python实现简单层次聚类算法,以及可视化,供大家参考,具体内容如下

基本的算法思路就是:把当前组间距离最小的两组合并成一组。

算法的差异在算法如何确定组件的距离,一般有最大距离,最小距离,平均距离,马氏距离等等。

代码如下:

  • ?
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25
  • 26
  • 27
  • 28
  • 29
  • 30
  • 31
  • 32
  • 33
  • 34
  • 35
  • 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
  • 运行的效果就是迭代一次,组数就会少一次,调用画图方法,同一组的数据被显示为一个颜色。

    以上就是本文的全部内容,希望对大家的学习有所帮助,也希望大家多多支持开心学习网。

    原文链接:https://blog.csdn.net/york1996/article/details/86652330

    您可能感兴趣