cluster.cpp

无监督聚类的VC源码。。数据挖掘中的一种经典聚类算法

#include <stdio.h>
#include <math.h>
#include <stdlib.h>
typedef  float Type;
#define  Max 100
int d;
float dist;


int count=0;

typedef struct Node
{
  Type *item;
  Node* next;
}LNode,*Linklist;

struct Cluster  //类的结构
{
  int size;
  float *center;
  Linklist member;
}*cluster;

int ReadData(FILE *fp, Type  *data) //从文件中读取数据
{
   for(int i=0;i<d+1;i++)
	   if(feof(fp)) return 0;
       else  fscanf(fp,"%f",&data[i]);
	   return 1;
}



void CreateList(Linklist &L)
{
  L=(Linklist)malloc(sizeof(LNode));
  L->next=NULL;
};




void insertList(Linklist &L,Type *data) //将一值插入列表
{
    Linklist p;
	int i;
    p=(Linklist)malloc(sizeof(LNode));
  	p->item=(Type *)malloc(sizeof(Type)*(d+1));
	for(i=0;i<d+1;i++)
         p->item[i]=data[i];
    p->next=L->next;
	L->next=p;
};


void initcluster(int i)
{
  cluster[i].size=0;
  CreateList(cluster[i].member);
  cluster[i].center=(float *)malloc(sizeof(float)*d);
}

float distance(Type* a,Type* b)    //两个值之间的欧几里得距离
{
   float dist=0.0;
   int i;
   for(i=0;i<d;i++)
      dist+=(a[i]-b[i])*(a[i]-b[i]);
  return (float)sqrt(dist);

}

float findclosedist(Type *temp,int q,int &t)
{
float MinDist, des;
int i=0;
MinDist=65536.0;
 for(i=0;i<q;i++)
 {
	 des=distance(temp,cluster[i].center);
      if (des<MinDist) {
           MinDist=des;
           t=i;
		} 
 }
 return MinDist;
}





void main()
{
  FILE *fp;
  Type *temp;
  fp=fopen("parameter.txt","r");
  fscanf(fp,"%d %f",&d,&dist);
  fclose(fp);
  temp=(Type *)malloc(sizeof(Type)*(d+1));
  int i,j,CID;
 
  cluster=(Cluster *)malloc(sizeof(Cluster) * Max);

fp=fopen("mydataset.txt","r"); 
ReadData(fp,temp);
initcluster(count);
for(i=0;i<d;i++)
cluster[count].center[i]=temp[i];
insertList(cluster[count].member,temp);
count++;
	 

while(ReadData(fp,temp))
{

if(findclosedist(temp,count,CID)<=dist) 
{ for(i=0;i<d;i++)
   cluster[CID].center[i]=(cluster[CID].center[i]*cluster[CID].size+temp[i])/(cluster[CID].size+1);
 insertList(cluster[CID].member,temp);
 cluster[CID].size++;
}
else
{
  
  initcluster(count);
  for(i=0;i<d;i++)
   cluster[count].center[i]=temp[i];
  insertList(cluster[count].member,temp);
  cluster[count].size++;
  count++;
}


};

 fclose(fp);
 fp=fopen("cluster.txt","w");
 printf("%d\n",count); 
 Linklist p;
  for(i=0;i<count;i++)
  {
	  fprintf(fp,"Center %d:",i+1);
	  for(j=0;j<d;j++)
	  fprintf(fp,"%f  ",cluster[i].center[j]);
	  fprintf(fp,"\n");
	  p=cluster[i].member->next;
   while(p)
   {
     for(j=0;j<d;j++)		 
		 fprintf(fp,"%f  ",p->item[j]);
	 fprintf(fp,"%d\n",(int)p->item[d]);
	 p=p->next;
   }
   fprintf(fp,"\n");
  
  }
fclose(fp);

}