verify-main.c

Ripper 分类算法

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


/******************************************************************************/
char *Program="verify";
char *Help_str[] = {
    "syntax: verify [options] [stem]",
    "   do a consistency check on a dataset",
    "",
    "options are:",
    "  -s:       read from stdinput",
    "  -v#:      set verbosity",
    NULL
};


main(argc,argv)
int argc;
char *argv[];
{
    vec_t *data;
    char *stem;
    int use_stdin;
    int i,j,o;
    ex_count_t p,n;
    example_t *exi;
    symbol_t *missing_sym;
    BOOL *missing_count;
    int minclass;
    double old_info;
    
    /* defaults */
    use_stdin = FALSE;
    set_trace_level(SUMM);

    while ((o=getopt(argc,argv,"stv:h"))!=EOF) {
	switch (o) {
	  case 's':
	    use_stdin = TRUE;
	    break;
	  case 'v':
	    set_trace_level(atoi(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 {
	data = ld_data(NULL);
    }

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

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

    trace(SUMM) {
	printf("= %d examples %d features %d values\n", 
	       vmax(data),n_fields(),n_symbolic_values());
	/* count class distribution */
	printf("\n= num\t  pct\tclass\n");
	for (i=0; i<vmax(Classes); i++) {
	    count_class_freq(vref(atom_t,Classes,i)->nom,data,&p,&n);
	    printf("%6.2f\t%6.2f\t",p,(p*100.0)/(p+n));
	    print_atom(vref(atom_t,Classes,i));
	    printf("\n");
	}
	fflush(stdout);
    }
    trace(LONG) {
	missing_count = newmem(n_fields(),int); 
	for (i=0; i<vmax(data); i++) {
	    exi = vref(example_t,data,i);
	    for (j=0; j<n_fields(); j++) {
		if (vref(aval_t,exi->inst,j)->kind == MISSING_VALUE) {
		    missing_count[j]++;
		}
	    }
	}
	printf("\n= attrib             type          #miss  #values\n");
	for (i=0; i<n_fields(); i++) {
	    printf("%-20.20s %6.6s\t",
		   field_name(i),
		   symbolic_field(i)? "symbol":(set_field(i)? "set" : "number"));
	    if (missing_count[i]==0) printf("%6s","no");
	    else printf("%6d",missing_count[i]);
	    printf("%8d",n_different_field_values(i,data));
	    printf("\n");
	}/*for i*/
    }
}