corrupt-main.c

Ripper 分类算法

/*****************************************************************************
 * corrupt-main.c --- main routine to add class noise to a dataset
 ****************************************************************************/


#include <stdio.h>
#include "ripper.h"
#include "mdb.h"
#include "protos.h"

/******************************************************************************/
char *Program="corrupt";
char *Help_str[] = {
    "syntax: corrupt [options] [stem]",
    "   adds class noise to data",     
    "",
    "options are:",
    "  -s:       read from stdin (not stem.data)",
    "  -p#:      class noise rate in pct (default 20)",  
    "  -R:       randomize operation",
    NULL
};


main(argc,argv)
int argc;
char *argv[];
{
    int i,j;
    example_t *exi;
    int o; 
    vec_t *data;
    char *stem;
    int use_stdin;
    int pct;
    double frac;
    
    pct = 20;
    while ((o=getopt(argc,argv,"stp:Rh"))!=EOF) {
	switch (o) {
	  case 'R':
	    randomize();
	    break;
	  case 's':
	    use_stdin = TRUE;
	    break;
	  case 'p':
	    frac = atof(optarg);
	    if (frac<1.0 && frac>0.0) pct = 100*frac;
	    else pct=frac;
	    break;
	  case 'h':
	  case '?':
	  default: 
	    give_help();
	    if (o=='h') exit(0);
	    else fatal("option not implemented");
	}
    }
    if (optind<argc) {
	stem = argv[optind++];
	ld_names(add_ext(stem,".names"));
	if (use_stdin) data = ld_data(NULL);
	else data = ld_data(add_ext(stem,".data"));
    } else {
	/* no stem */
	stem = "stdin";
	data = ld_data(NULL);
    }
    if (data==NULL || vmax(data)==0) fatal("no data!");

    if (optind<argc) {
	warning("not all arguments were used: %s ...",argv[optind]);
    }

    if (pct<0 || pct>=100) {
	fatal("p must be between 0 and 100");
    }
    for (i=0; i<vmax(data); i++) {
	exi = vref(example_t,data,i); 	
	if (random()%100 < pct) {
	    j = random()%vmax(Classes);
	    exi->lab.nom = vref(atom_t,Classes,j)->nom;
	}
    }
    fprint_data(stdout,data);
}