Implement a simple approach for k-means/ k-medoids clustering using C++

PROGRAM

filename: a6.cpp

#include<iostream>
#include<stdio.h>
#include<time.h>
#include<stdlib.h>
using namespace std;

class kmeans
{

struct points
{
 float x;
 float y;
};
public:
struct points p[8];
struct points c[3];
void input();
int km();
float distance(float a,float b,float x,float y);
};
int tally[8][10];
int count;
int n;

void kmeans::input()
{
 p[0].x=2;p[0].y=2;
 p[1].x=1;p[1].y=14;
 p[2].x=10;p[2].y=7;
 p[3].x=1;p[3].y=11;
 p[4].x=3;p[4].y=4;
 p[5].x=11;p[5].y=8;
 p[6].x=4;p[6].y=3;
 p[7].x=12;p[7].y=2;
int i,random=0;
srand(time(NULL));
for(i=0;i<3;i++)
{
 random=(rand())%7;
 c[i].x=p[random].x;
 c[i].y=p[random].y;
}
}

float kmeans::distance(float a,float b,float x,float y)
{
float t1=a-x;
float t2=b-y;
if(t1<0)
 t1=-t1;
if(t2<0)
 t2=-t2;
return (t1+t2);
}

int kmeans::km()
{
int i,j;
float table[8][4];
for(i=0;i<n;i++)
{
 table[i][0]=distance(p[i].x,p[i].y,c[0].x,c[0].y);
 table[i][1]=distance(p[i].x,p[i].y,c[1].x,c[1].y);
 table[i][2]=distance(p[i].x,p[i].y,c[2].x,c[2].y);
 if(table[i][0]<=table[i][1] && table[i][0]<=table[i][2])
 {
  tally[i][count]=0;
  table[i][3]=table[i][0];
 }
 else if(table[i][1]<=table[i][0] && table[i][1]<=table[i][2])
 {
  tally[i][count]=1;
  table[i][3]=table[i][1];
 }
 else if(table[i][2]<=table[i][0] && table[i][2]<=table[i][1])
 {
  tally[i][count]=2;
  table[i][3]=table[i][2];
 }
}
cout<<"TABLE: "<<count+1<<endl;
cout<<"Centroids are: ("<<c[0].x<<","<<c[0].y<<"), ("<<c[1].x<<","<<c[1].y<<"), ("<<c[2].x<<","<<c[2].y<<")"<<endl;

cout<<"c1\tc2\tc3\tmin"<<endl;
cout<<"-------------------------"<<endl;
for(i=0;i<n;i++)
{
 for(j=0;j<4;j++)
 {
 cout<<table[i][j]<<"\t";
 }
cout<<endl;
}
cout<<"-------------------------"<<endl;

float c0=0,c1=0,c2=0,c0x=0,c0y=0,c1x=0,c1y=0,c2x=0,c2y=0;
for(i=0;i<n;i++)
{
if(tally[i][count]==0)
{
c0x+=p[i].x;
c0y+=p[i].y;
c0++;
}
else if(tally[i][count]==1)
{
c1x+=p[i].x;
c1y+=p[i].y;
c1++;
}
else if(tally[i][count]==2)
{
c2x+=p[i].x;
c2y+=p[i].y;
c2++;
}
}
c[0].x=c0x/c0;
c[0].y=c0y/c0;
c[1].x=c1x/c1;
c[1].y=c1y/c1;
c[2].x=c2x/c2;
c[2].y=c2y/c2;

int flag=0;
if(count!=0)
{
for(i=0;i<n;i++)
{
 if(tally[i][count]==tally[i][count-1])
   flag++;
}
}
count++;
return flag;
}

int main()
{
kmeans k;
k.input();
n=8;
count=0;
int flag=k.km();
while(flag!=8)
 flag=k.km();
if(flag==8)
 cout<<"Centroids found"<<endl;
for(int i=0;i<n;i++)
{
 for(int j=0;j<count;j++)
  cout<<tally[i][j]<<"  ";
 cout<<endl;
}
return 0;
}


OUTPUT

amodi@localhost ~]$ g++ a6.cpp
[amodi@localhost ~]$ ./a.out
TABLE: 1
Centroids are: (11,8), (3,4), (10,7)
c1 c2 c3 min
-------------------------
15 3 13 3
16 12 16 12
2 10 0 0
13 9 13 9
12 0 10 0
0 12 2 0
12 2 10 2
7 11 7 7
-------------------------
TABLE: 2
Centroids are: (11.5,5), (2.2,6.8), (10,7)
c1 c2 c3 min
-------------------------
12.5 5 13 5
19.5 8.4 16 8.4
3.5 8 0 0
16.5 5.4 13 5.4
9.5 3.6 10 3.6
3.5 10 2 2
9.5 5.6 10 5.6
3.5 14.6 7 3.5
-------------------------
TABLE: 3
Centroids are: (12,2), (2.2,6.8), (10.5,7.5)
c1 c2 c3 min
-------------------------
10 5 14 5
23 8.4 16 8.4
7 8 1 1
20 5.4 13 5.4
11 3.6 11 3.6
7 10 1 1
9 5.6 11 5.6
0 14.6 7 0
-------------------------
Centroids found
1  1  1  
1  1  1  
2  2  2  
1  1  1  
1  1  1  
0  2  2  
1  1  1  
0  0  0