熵聚类程序辅导、讲解熵聚类编程

惩罚度1=0

药物出现次数1=10

from     math  import  log  #求对数

import  xlrd,xlwt

from  numpy import  mat

from   pprint  import   pprint

import networkx as nx  # 复杂网络需要用的包

import matplotlib.pyplot as plt  # 画图用

plt.rcParams['font.sans-serif'] = ['SimHei'] #设置字体为中文字体

#计算熵

def unionShannonEnt(x1,x2,n=惩罚度1):

   numEntries = int(len(x1))

   # print(numEntries)

   X00=float(0);X11=float(0);X10=float(0);X01=float(0);

   #（以下五行）为所有可能分类创建字典

   # print(x1.sum(axis=0),x2.sum(axis=0))

   for i in range(numEntries):

       if  x1[i,0]==0 and x2[i,0]==0:X00+=1

       elif x1[i,0]==1 and x2[i,0]==1: X11+=1

       elif x1[i,0] == 0 and x2[i,0]:X10 += 1

       else:X01 += 1

   # print(X00,X01,X10,X11)

   numEntries=float(numEntries)

   if  (X00+X01)/numEntries>0:

       a1=-(X00+X01)/numEntries*(log((X00+X01)/numEntries,10))

   else: a1=0

   if  (X10+X11)/numEntries>0:

       b1 =  -(X10+X11)/numEntries*(log((X10+X11)/numEntries,10))

   else:b1=0

   if (X10+X00)/numEntries>0:

       c1 = -(X10+X00)/numEntries*log((X10+X00)/numEntries,10)

   else: c1=0

   if  (X01+X11)/numEntries>0:

       d1  = -(X01+X11)/numEntries*log((X01+X11)/numEntries,10)

   else:d1=0

   H_x1=a1+b1

   H_x2=c1+d1

   if  X00/numEntries>0:

       a=-X00/numEntries*log(X00/numEntries,10)

   else: a=0

   if X11/numEntries >0:

       b = -X11/numEntries * log(X11 / numEntries, 10)

   else:b=0

   if  X01/numEntries>0:

       c = -X01/numEntries*log(X01/numEntries,10)

   else: c=0

   if  X10/numEntries>0:

       d  = -X10/numEntries*log(X10/numEntries,10)

   else:d=0

   Hx1_x2=a+b+c+d

   if X11>n:

       unionEnt=(H_x1+H_x2-Hx1_x2)/(H_x1*H_x2)**0.5

   else:

       if  H_x1*H_x2 >0:

           unionEnt =(H_x1 + H_x2 - 2*Hx1_x2)/(H_x1 * H_x2) ** 0.5

       else:unionEnt =0.1

   return unionEnt

#熵就是分类为0或者1的数目

def  Readexcel(table,sheet):

   data=xlrd.open_workbook(table)

   try:

       # table=data.sheets()[2]  #通过索引顺序获取

       # # table = data.sheet_by_index()  # 通过索引顺序获取

       table = data.sheet_by_name(sheet)   # 通过名称获取d

   except:

       print("no sheet  in %s named  sheet1"%data)

   print(table.nrows,table.ncols)   #读取数据的行和列

   dataset=[];labels=[]

   for  n  in  range(table.nrows):

       col=[]

       for  c in range(table.ncols):

               col.append(table.cell(n,c).value)

       if  n>0 :

           dataset.append(col[:])

       if n==0:

           labels.append(col[:])

   return dataset,labels

#读取数据

xMat,labels=Readexcel(r"C:\Users\Administrator\Desktop\方剂.xlsx",u'data')

xMat=mat(xMat)

# print(xMat)

print(labels)

z=unionShannonEnt(xMat[:,1],xMat[:,2],n=5)

print(z)

result_2=[]

set_result2={}

res=float(0)

for  m  in range(xMat.shape[1]):

   for n  in range(xMat.shape[1]):

       if m <n and xMat[:,m].sum(axis=0) >药物出现次数1 and  xMat[:,n].sum(axis=0)>药物出现次数1: #相关度

           # print(m,n)

           res = float(unionShannonEnt(xMat[:,m], xMat[:,n], n=0))

           result_2.append([labels[0][m],labels[0][n],res])

           set_result2[frozenset((labels[0][m],labels[0][n]))]=res

print(result_2[0]) #

result_2.sort(key=lambda x: x[2], reverse=True)   #数据降序排列.

print(result_2)#药物关联度

#生成三个类别、四个类别的组合

result_3=[]

for  p in  range(len(result_2)):

   for  q    in  range(len(result_2)):

       if  p<q  and result_2[p][2]>0.13 and result_2[q][2]>0.13  :

           result_3.append(list(set(result_2[p][0:2])|set(result_2[q][0:2])))

print(result_3)

# 生成绘制网络图的数据

G = nx.Graph()  # 创建无向图

#添加节点

for  z1  in result_3:

   for  z2  in  z1  :

       for  z3 in z1  :

               if  frozenset((z2,z3))  in set_result2.keys() :

                   if   set_result2[frozenset((z2,z3))] >0.01:

                       G.add_edge(z2,z3)

# 创建画布，绘制和展示图形

plt.figure()  # 创建一幅图

nx.draw(G, node_color='white',with_labels=True, node_size=300,font_size=11,font_color='black')

# plt.figure()  # 创建一幅图

# nx.draw(G, node_color='black',with_labels=False, node_size=30)

plt.show()

#导出数据

import   xlwt

book = xlwt.Workbook(encoding='utf-8', style_compression=0)

sheet = book.add_sheet('关联', cell_overwrite_ok=True)

sheet2 = book.add_sheet('聚类', cell_overwrite_ok=True)

n=0

for  x in  result_2:

   # print(x)

   sheet.write(n, 1, str(x))

   n+=1

b=0

for  y  in  result_3:

   sheet2.write(b, 1, str(y))

   b += 1

book.save(r'd:\聚类.csv')  #保存结果