cmeans11.m

人工智能中比较经典的c-means算法的实现。在MATLAB中

clear;
clc;
clf;
%CMEANS Summary of this function goes here
%   Detailed explanation goes here
[a1,a2,a3,a4]=textread('男女学生数据\StudentsData.txt','%f%f%f%f');
A=[a1,a2,a3,a4];
Member=[a1];
TestSet.X=[a3,a4];
c.num=0;
c1=TestSet.X(1,:);
%c2=TestSet.X(2,:);
N=size(TestSet.X,1);
d=size(TestSet.X,2);
dis=0;
for i=1:N
 dd=0;
   for j=1:d
       dd=dd+(c1(1,j)-TestSet.X(i,j))^2;
   end
   if dd>dis
       dis=dd;
       c2=TestSet.X(i,:);
       t=i;
   end
end
%Member(t);

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%初次划分
t1=0;
t2=0;
for i=1:N
      d1=0;
      d2=0;
   for j=1:d
       d1=d1+(c1(1,j)-TestSet.X(i,j))^2;
       d2=d2+(c2(1,j)-TestSet.X(i,j))^2;
   end
   if d1>d2
        t2=t2+1;
        Member2(t2,1)=Member(i);
        setc2(t2,:)=TestSet.X(i,:);         
   else 
        t1=t1+1;
        Member1(t1,1)=Member(i);
        setc1(t1,:)=TestSet.X(i,:);         
   end
end
Member1;
Member2;
setc2;
setc1;
t1
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%聚类均值

%for i=1:d
    m1=sum(setc1)/t1;
%end
%for i=1:d
   m2=sum(setc2)/t2;
%end
m1
m2

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%准则函数值
J1=0;
J2=0;
for i=1:t1
    J1=J1+(setc1(i,:)-m1(1,:))*(setc1(i,:)-m1(1,:))';
end
for i=1:t2
    J2=J2+(setc2(i,:)-m2(1,:))*(setc2(i,:)-m2(1,:))';
end
Je=J1+J2
delta=1000;
Je1=0;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%循环划分
while delta>0.01

    nt2=1;
    nt1=1;
for i=1:t1
    p1=(setc1(i,:)-m1(1,:))*(setc1(i,:)-m1(1,:))'*t1/(t1-1);
    p2=(setc1(i,:)-m2(1,:))*(setc1(i,:)-m2(1,:))'*t2/(t2+1);
    
    if p1>p2
       tempsetc2(nt2,:)=setc1(i,:);
       nt2=nt2+1;
    else
       tempsetc1(nt1,:)=setc1(i,:);
       nt1=nt1+1;
    end
end
for i=1:t2
 p1=(setc2(i,:)-m1(1,:))*(setc2(i,:)-m1(1,:))'*t1/(t1+1);
 p2=(setc2(i,:)-m2(1,:))*(setc2(i,:)-m2(1,:))'*t2/(t2-1);
    if p1>p2
       tempsetc2(nt2,:)=setc2(i,:);
       nt2=nt2+1;
    else
       tempsetc1(nt1,:)=setc2(i,:);
       nt1=nt1+1; 
    end
end
setc1=tempsetc1;
setc2=tempsetc2;
t1=nt1-1;
t2=nt2-1;
clear tempsetc1*%=zeros(100,2);
clear tempsetc2*%=zeros(100,2);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%重新计算
 m1=sum(setc1)/t1;
 m2=sum(setc2)/t2;

for i=1:t1
    J1=J1+(setc1(i,:)-m1(1,:))*(setc1(i,:)-m1(1,:))';
end
for i=1:t2
    J2=J2+(setc2(i,:)-m2(1,:))*(setc2(i,:)-m2(1,:))';
end
Je1=J1+J2;
delta=abs(Je-Je1)/Je;
Je=Je1;
end
setc1
setc2
Je1

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%求错分率
t3=0;
for i=1:t1
    for j=1:N
      if (setc1(i,1)==a3(j)&&setc1(i,2)==a4(j))
         if a2(j)==-1
            t3=t3+1;
            setc3(t3,:)=setc1(i,:);
         end
      end
    end
end
for i=1:t2
    for j=1:N
      if (setc2(i,1)==a3(j)&&setc2(i,2)==a4(j))
          if a2(j)==1
            t3=t3+1;
            setc3(t3,:)=setc2(i,:);
          end
      end
    end
   
end
setc3
e=t3/N
plot(setc1(:,1),setc1(:,2),'+',setc2(:,1),setc2(:,2),'o',setc3(:,1),setc3(:,2),'x')