partition-main.c

Ripper 分类算法

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


/******************************************************************************/
char *Program="partition";
char *Help_str[] = {
    "syntax: partition [options] [stem]",
    "   choose a random subset of a dataset",
    "",
    "options are:",
    "  -s:       read from stdin (not stem.data)",
    "  -m#:      choose # examples",  
    "  -p#:      choose a percent # of the examples",  
    "  -r<str>:  put remaining examples in file <str>",  
    "  -R:       randomize operation",
    NULL
};


main(argc,argv)
int argc;
char *argv[];
{
    int m,i,j,ndel;
    int o; 
    vec_t *data;
    char *stem;
    FILE *rfp;
    BOOL use_stdin;
    int pct;
    double frac;
    
    m = 0;
    rfp = NULL;
    pct = 50;
    use_stdin = FALSE;
    set_trace_level(NONE);
    while ((o=getopt(argc,argv,"stm:p:r:Rh"))!=EOF) {
	switch (o) {
	  case 'R':
	    randomize();
	    break;
	  case 's':
	    use_stdin = TRUE;
	    break;
	  case 'm':
	    m = atoi(optarg);
	    break;
	  case 'p':
	    frac = atof(optarg);
	    if (frac<=1.0 && frac>=0.0) pct = 100*frac;
	    else pct=frac;
	    break;
	  case 'r':
	    rfp = fopen(optarg,"w");
	    if (rfp==NULL) fatal("can't open file %s for output",optarg);
	    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 (optind<argc) {
	warning("not all arguments were used: %s ...",argv[optind]);
    }

    if (data==NULL || vmax(data)==0) fatal("no data!");

    if (m!=0) { 
	/* remove m random examples */
	if (m<=0 || m >= vmax(data)) {
	    fatal("m must be >=0 and <= %d(was %d)\n",vmax(data),m);
	}
	stratify_and_shuffle_data(data,1);
	ndel = vmax(data)-m;
	
	for (i=0; i<ndel; i++) {
	    j = random()%vmax(data);
	    if (rfp) fprint_example(rfp,vref(example_t,data,j));
	    swap_out_example(data,j);
	}
    } else {
	if (pct<=0 || pct>=100) {
	    fatal("p must be between 0 and 100");
	}
	for (i=0; i<vmax(data); i++) {
	    if (random()%100 > pct) {
		if (rfp) fprint_example(rfp,vref(example_t,data,i));
		swap_out_example(data,i);
	    }
	}
    }
    fprint_data(stdout,data);
}