思路

1. 輸入數據：x,y分別對應橫縱座標
2. 確定分類數k
3. 給出初始的k個聚類中心
4. 計算數據到k箇中心的聚類，分別都歸類到距離最近的中心
5. 歸類後的k組數據取中心點作為新的聚類中心
6. 迭代直到收斂

具體代碼

<code>clc
clear all
close all
format bank;
x=[1 1.5 3 5 3.5 4.5 3.5];
y=[1 2 4 7 5 5 4.5];
k=2;
m1_newx=1;
m1_newy=1;
m2_newx=5;
m2_newy=7;
temp=0;
k1x=[];
k1y=[];
k2x=[];
k2y=[];
while(temp==0)
    cx=[m1_newx m2_newx];
    m1_oldx=cx(1);%m1_oldx=m1_newx
    m2_oldx=cx(2);%m2_oldx=m2_newx
    cy=[m1_newy m2_newy];
    m1_oldy=cy(1);%m1_oldy=m1_newy
    m2_oldy=cy(2);%m2_oldy=m2_newy
    for i=1:length(x)
        xx1=(x(i)-m1_oldx)^2;
        yy1=(y(i)-m1_oldy)^2;
        g=xx1+yy1;
        g1=sqrt(g);
        xx2=(x(i)-m2_oldx)^2;
        yy2=(y(i)-m2_oldy)^2;
        g=xx2+yy2;
        g2=sqrt(g);
        if(g1>g2)
            k2x=[k2x x(i)];
            k2y=[k2y y(i)];
        else
            k1x=[k1x x(i)];
            k1y=[k1y y(i)];
        end
    end
    m1_newx=mean(k1x);
    m2_newx =mean(k2x);
    m1_newy=mean(k1y);
    m2_newy =mean(k2y);
    if(m1_newx ~=m1_oldx && m2_newx ~=m2_oldx && m1_newy ~=m1_oldy && m2_newy ~=m2_oldy)
        k1x=[];
        k2x=[];
        k2y=[];
        k1y=[];
    else
        temp=1;
    end
end
for i=1:length(k1x)
    t=0:0.01:2*pi;
    x=k1x(i)+0.1*cos(t);
    y=k1y(i)+0.1*sin(t);
    fill(x,y,'r');
    hold on;
    axis square;
    grid on;
end
hold on;
for i=1:length(k2x)
    t=0:0.01:2*pi;
    x=k2x(i)+0.1*cos(t);
    y=k2y(i)+0.1*sin(t);
    fill(x,y,'g');
    axis square;
    hold on;
    grid on;
end/<code>

結果展示

閱讀更多 學本無止境 的文章

關鍵字: 聚類 中心 Square