sbg.c~.txt

C语言的朴素贝叶斯分类器代码

/****************************************************************************
* The algorithm implemented here can be found in the book titled            *
*    FEATURE SELECTION FOR KNOWLEDGE DISCOVERY by Huan Liu &                *
* Hiroshi Motoda published by Kluwer in 1998, ISBN 0-7923-8198-X            *
*                                                                           *
* The program developed by:                                                 *
*                                                                           *
*                       Md. Farhad Hussain                                  *
*                           DISCS, NUS                                      *
*                    Kent Ridge, S'pore 119260                              *
*                  email: [e-mail]                            *
*                          [e-mail]                             *
*****************************************************************************/

#include "header.h"

/*----------------------------Global Definitions-----------------------------*/
long *S, *F, *L;
long indx;
/*-------------------------External Definitions------------------------------*/
extern void ReadData(char*);
extern long COL, ROW;
/*---------------------------------------------------------------------------*/
void InitSBG(){
long i;

  indx = 0;
  S = (long*)malloc(sizeof(long) * (COL - 1));
  SUCCESS(S);
  F = (long*)malloc(sizeof(long) * (COL - 1));
  SUCCESS(F);
  L = (long*)malloc(sizeof(long) * (COL - 1));
  SUCCESS(L);

  for( i = 0; i < COL - 1; i++ ){
	S[i] = i;
	F[i] = EMPTY;
	L[i] = EMPTY;
  }
}
/*---------------------------------------------------------------------------*/
void ConsiderRemovingFeature(long feature, char *datafile){
char command[256];
long i;

  system("rm -f Expr.data");
  sprintf(command,"cut -d, -f"); 
  for(i = 0; i < COL - 1; i++ ){
	if( S[i] != feature && S[i] != EMPTY )
		sprintf(command,"%s%ld,", command, S[i]+1);
  }
  sprintf(command, "%s%ld %s > Expr.data", command, COL, datafile); 
  system(command);
}
/*---------------------------------------------------------------------------*/
float CalculateError(){
FILE *fp;
char buffer[256];

  system("rm -f nbc.result");
  system("echo Expr.data Expr.data | nbc > nbc.result");
  fp = fopen("nbc.result","r");
  fgets(buffer, 256, fp);
  while( strstr(buffer, "%") == NULL )
	fgets(buffer, 256, fp);
  fclose(fp);
  buffer[strlen(buffer) - 1] = 0;
  return atof( strstr(buffer, "=") + 1);
}
/*---------------------------------------------------------------------------*/
long RemoveNextFeature(){
long i;

  for( i = 0; i < COL - 1; i++ ){
	if( F[i] == EMPTY ){
		F[i] = i;
		return i;
	}
  }
  return EMPTY; 
}
/*---------------------------------------------------------------------------*/
void RemoveFeature(long feature){
long i;

  for( i = 0; S[i] != feature; i++ );
  S[i] = EMPTY;
  
  printf("\n");
  for( i = 0; i < COL - 1; i++ ){
	if( S[i] == EMPTY )
		F[i] = i;
	else{
		printf("A%ld ", S[i]);
		F[i] = EMPTY;
	}
  } 

  L[indx++] = feature;
}
/*---------------------------------------------------------------------------*/
void RunSBG(char* datafile){
long i, BestFeature, feature;
float error, tempError; 

  for( i = 0; i < COL - 2; i ++ ){
	error = 0;
	while((feature = RemoveNextFeature()) != EMPTY){
		ConsiderRemovingFeature( feature, datafile);
		if( (tempError = CalculateError()) > error ){
			error = tempError;
			BestFeature = feature;
  		}
  	}
	RemoveFeature( BestFeature );
  }
  printf("\n\n");
  for( i = 0; i < COL - 2; i++ )
	printf("A%ld ", L[i]);
  for( i = 0; S[i] == EMPTY; i++ );
	printf("A%ld", i);
}
/*---------------------------------------------------------------------------*/
main(int argc, char* argv[]){

  ReadData(argv[1]);
  InitSBG();
  RunSBG(argv[1]);
  printf("\n");
}