summary.c~

Solaris环境下的数据挖掘算法：birch聚类算法。该算法适用于对大量数据的挖掘。

#include <iostream.h>
#include <fstream.h>
#include <math.h>
#include "vector.h"

#define NUMCLUSTERS 100
#define NUMATTRS    2

void Summary(char *labelfile, char *inputfile) 
{ 
  ifstream ifile;
  ifstream labFile;
  ofstream ofile;
  Vector summary[NUMCLUSTERS];
  Vector temp;
  temp.Init(NUMATTRS);
 
  int		 z=-1;
  int		 number=0;
  int		 numvecs[NUMCLUSTERS];
  double	 metric = 0.0;
  
  ifile.open(inputfile);
  labFile.open(labelfile);
  
  for (int i=0; i < NUMCLUSTERS; i++){
    summary[i].Init(NUMATTRS);
  }
  while(labFile.peek() != EOF) {
    for (int i=0; i < NUMATTRS; i++){
      ifile >> temp.value[i];
    }
    labFile >> z;
    summary[z] += temp;
    numvecs[z]++;
    number++;
  }
  cout << " NUMBER :" << number << endl;
  for (int i=0; i < NUMCLUSTERS; i++){
    summary[z] /= numvecs[z];
    ofile << numvecs[z] << "   ";
    ofile << summary[z] << endl;
  }
  ifile.close();
  labFile.close();
  ifile.open(inputfile);
  labFile.open(labelfile);
  
  while (ifile.peek() != EOF){
    for (int i=0; i < NUMATTRS; i++){
      ifile >> temp.value[i];
    }
    labFile >> z;
    metric += sqrt( summary[z] || temp);
  }
  ofile << " The cluster metric is " << metric << endl;
}

main(int argc, char* argv) 
{
  if (argc < 3)
    {
      cout << " Usage: summary <label filename> <data filename>" << endl;
      exit(1);
    }
  Summary(&argv[1], &argv[2]);
}