dtx.c

it is tree Dtree algorithm. so see it. it is in c++

/*--------------------------------------------------------------------*/

int main (int argc, char* argv[])
{                               /* --- main function */
  int    i, k = 0, f;           /* loop variables, counter */
  char   *s;                    /* to traverse options */
  char   **optarg = NULL;       /* option argument */
  char   *fn_hdr  = NULL;       /* name of table header file */
  char   *fn_tab  = NULL;       /* name of table file */
  char   *fn_dt   = NULL;       /* name of dec./reg. tree file */
  char   *fn_out  = NULL;       /* name of output file */
  char   *blanks  = NULL;       /* blanks */
  char   *fldseps = NULL;       /* field  separators */
  char   *recseps = NULL;       /* record separators */
  char   *uvchars = NULL;       /* unknown value characters */
  int    inflags  = 0;          /* table file read  flags */
  int    outflags = AS_ATT;     /* table file write flags */
  int    tplcnt   = 0;          /* number of tuples */
  double tplwgt   = 0.0;        /* weight of tuples */
  double errcnt   = 0.0;        /* number of misclassifications */
  int    attid;                 /* loop variable for attributes */
  float  wgt;                   /* tuple/instantiation weight */
  INST   *inst;                 /* instance of target attribute */
  double diff;                  /* difference to true value */
  TFSERR *err;                  /* error information */

  prgname = argv[0];            /* get program name for error msgs. */

  /* --- print startup/usage message --- */
  if (argc > 1) {               /* if arguments are given */
    fprintf(stderr, "%s - %s\n", argv[0], DESCRIPTION);
    fprintf(stderr, VERSION); } /* print a startup message */
  else {                        /* if no argument is given */
    printf("usage: %s [options] dtfile "
                     "[-d|-h hdrfile] tabfile [outfile]\n", argv[0]);
    printf("%s\n", DESCRIPTION);
    printf("%s\n", VERSION);
    printf("-p#      prediction field name "
                    "(default: %s)\n", res.n_pred);
    printf("-s#      support    field name "
                    "(default: no support    field)\n");
    printf("-c#      confidence field name "
                    "(default: no confidence field)\n");
    printf("-a       align fields (default: do not align)\n");
    printf("-w       do not write field names to output file\n");
    printf("-b/f/r#  blank characters, field and record separators\n"
           "         (default: \" \\t\\r\", \" \\t\", \"\\n\")\n");
    printf("-u#      unknown value characters (default: \"?\")\n");
    printf("-n       number of tuple occurrences in last field\n");
    printf("dtfile   file containing decision/regression "
                    "tree description\n");
    printf("-d       use default header "
                    "(field names = field numbers)\n");
    printf("-h       read table header (field names) from hdrfile\n");
    printf("hdrfile  file containing table header (field names)\n");
    printf("tabfile  table file to read "
                    "(field names in first record)\n");
    printf("outfile  file to write output table to (optional)\n");
    return 0;                   /* print a usage message */
  }                             /* and abort the program */

  /* --- evaluate arguments --- */
  for (i = 1; i < argc; i++) {  /* traverse arguments */
    s = argv[i];                /* get option argument */
    if (optarg) { *optarg = s; optarg = NULL; continue; }
    if ((*s == '-') && *++s) {  /* -- if argument is an option */
      while (1) {               /* traverse characters */
        switch (*s++) {         /* evaluate option */
          case 'p': optarg    = &res.n_pred; break;
          case 's': optarg    = &res.n_supp; break;
          case 'c': optarg    = &res.n_conf; break;
          case 'a': outflags |=  AS_ALIGN;   break;
          case 'w': outflags &= ~AS_ATT;     break;
          case 'b': optarg    = &blanks;     break;
          case 'f': optarg    = &fldseps;    break;
          case 'r': optarg    = &recseps;    break;
          case 'u': optarg    = &uvchars;    break;
          case 'n': outflags |= AS_WEIGHT;
                    inflags  |= AS_WEIGHT;   break;
          case 'd': inflags  |= AS_DFLT;     break;
          case 'h': optarg = &fn_hdr;        break;
          default : error(E_OPTION, *--s);   break;
        }                       /* set option variables */
        if (!*s) break;         /* if at end of string, abort loop */
        if (optarg) { *optarg = s; optarg = NULL; break; }
      } }                       /* get option argument */
    else {                      /* if argument is no option */
      switch (k++) {            /* evaluate non-option */
        case  0: fn_dt  = s;      break;
        case  1: fn_tab = s;      break;
        case  2: fn_out = s;      break;
        default: error(E_ARGCNT); break;
      }                         /* note filenames */
    }
  }
  if (optarg) error(E_OPTARG);  /* check option argument */
  if ((k < 2) || (k > 3)) error(E_ARGCNT);
  if (fn_hdr && (strcmp(fn_hdr, "-") == 0))
    fn_hdr = "";                /* convert "-" to "" */
  i = (!fn_dt || !*fn_dt) ? 1 : 0;
  if  (fn_tab && !*fn_tab) i++;
  if  (fn_hdr && !*fn_hdr) i++; /* check assignments of stdin: */
  if (i > 1) error(E_STDIN);    /* stdin must not be used twice */
  if (fn_hdr)                   /* set the header file flag */
    inflags = AS_ATT | (inflags & ~AS_DFLT);
  if ((outflags & AS_ATT) && (outflags & AS_ALIGN))
    outflags |= AS_ALNHDR;      /* set align to header flag */

  /* --- read decision/regression tree --- */
  scan = sc_create(fn_dt);      /* create a scanner */
  if (!scan) error((!fn_dt || !*fn_dt) ? E_NOMEM : E_FOPEN, fn_dt);
  attset = as_create("domains", att_delete);
  if (!attset) error(E_NOMEM);  /* create an attribute set */
  fprintf(stderr, "\nreading %s ... ", sc_fname(scan));
  if ((sc_nexter(scan)   <  0)  /* start scanning (get first token) */
  ||  (as_parse(attset, scan, AT_ALL) != 0)
  ||  (as_attcnt(attset) <= 0)) /* parse attribute set */
    error(E_PARSE, sc_fname(scan));
  dtree = dt_parse(attset,scan);/* parse the dec./reg. tree */
  if (!dtree || !sc_eof(scan)) error(E_PARSE, fn_dt);
  sc_delete(scan); scan = NULL; /* delete the scanner */
  fprintf(stderr, "[%d attribute(s)/%d node(s)/%d level(s)] done.\n",
          as_attcnt(attset), dt_size(dtree), dt_height(dtree));

  /* --- read table header --- */
  res.att  = as_att(attset, dt_trgid(dtree));
  res.type = att_type(res.att); /* get the target attribute */
  for (attid = as_attcnt(attset); --attid >= 0; )
    att_setmark(as_att(attset, attid), 1);
  att_setmark(res.att, 0);      /* mark all attribs. except the class */
  as_chars(attset, blanks, fldseps, recseps, uvchars);
  in = io_hdr(attset, fn_hdr, fn_tab, inflags|AS_MARKED, 1);
  if (!in) error(1);            /* read the table header */

  /* --- process tuples --- */
  if ((att_getmark(res.att) < 0)/* either the class must be present */
  &&  (k <= 2))                 /* or an output file must be written */
    error(E_CLASS, att_name(res.att), fn_tab);
  if (k > 2) {                  /* if to write an output table */
    if ((outflags & AS_ALIGN)   /* if to align output file */
    &&  (in != stdin)) {        /* and not to read from stdin */
      i = AS_INST | (inflags & ~(AS_ATT|AS_DFLT));
      while (as_read(attset, in, i) == 0);
      fclose(in);               /* determine the column widths */
      in = io_hdr(attset, fn_hdr, fn_tab, inflags|AS_MARKED, 1);
      if (!in) error(1);        /* reread the table header */
    }                           /* (necessary because of first tuple) */
    if (fn_out && *fn_out)      /* if an output file name is given, */
      out = fopen(fn_out, "w"); /* open the output file for writing */
    else {                      /* if no output file name is given, */
      out = stdout; fn_out = "<stdout>"; }       /* write to stdout */
    if (!out) error(E_FOPEN, fn_out);
    k = AS_MARKED|AS_INFO1|AS_RDORD|outflags;
    if (outflags & AS_ATT)      /* if to write table header */
      as_write(attset, out, k, infout);
    k = AS_INST|(k & ~AS_ATT);  /* write the attribute names */
  }                             /* to the output file */
  f = AS_INST | (inflags & ~(AS_ATT|AS_DFLT));
  i = ((inflags & AS_DFLT) && !(inflags & AS_ATT))
    ? 0 : as_read(attset, in, f);
  inst = att_inst(res.att);     /* get the instance of the target */
  while (i == 0) {              /* record read loop */
    dt_exec(dtree, NULL, &res.pred, &res.supp, &res.conf);
    wgt = as_getwgt(attset);    /* classify tuple */
    tplwgt += wgt; tplcnt++;    /* count tuple and sum its weight */
    if (res.type == AT_SYM) {   /* if the target att. is symbolic */
      if ((inst->i > UV_SYM) && (res.pred.i != inst->i))
        errcnt += wgt; }        /* count classification errors */
    else {                      /* if the target att. is numeric */
      if (res.type == AT_INT)   /* if it is integer-valued */
        diff = (inst->i > UV_INT) ? res.pred.f -inst->i : 0;
      else                      /* if it is float-valued */
        diff = (inst->f > UV_FLT) ? res.pred.f -inst->f : 0;
      errcnt += diff *diff;     /* compute the diff. to the true */
    }                           /* value and sum the squared errors */
    if (out && (as_write(attset, out, k, infout) != 0))
      error(E_FWRITE, fn_out);  /* write tuple to output file */
    i = as_read(attset, in, f); /* try to read the next record */
  }
  if (i < 0) {                  /* if an error occurred, */
    err = as_err(attset);       /* get the error information */
    tplcnt += (inflags & (AS_ATT|AS_DFLT)) ? 1 : 2;
    io_error(i, fn_tab, tplcnt, err->s, err->fld, err->exp);
    error(1);                   /* print an error message */
  }                             /* and abort the program */
  if (in != stdin) fclose(in);  /* close the table file and */
  in = NULL;                    /* clear the file variable */
  if (out && (out != stdout)) { /* if an output file exists, */
    i = fclose(out); out = NULL;/* close the output file */
    if (i) error(E_FWRITE, fn_out);
  }                             /* print a success message */
  fprintf(stderr, "[%d/%g tuple(s)] done.\n", tplcnt, tplwgt);
  if (att_getmark(res.att) >= 0) {   /* if the target is present */
    if (res.type != AT_SYM)     /* if the target att. is numeric */
      fprintf(stderr, "root of mean squared error: %g\n",
             (tplwgt > 0) ? sqrt(errcnt /tplwgt) : 0);
    else                        /* if the target att. is symbolic */
      fprintf(stderr, "%g error(s) (%.2f%%)\n", errcnt,
             (tplwgt > 0) ? 100*(errcnt /tplwgt) : 0);
  }                             /* print number of misclass. */

  /* --- clean up --- */
  #ifndef NDEBUG
  dt_delete(dtree, 1);          /* delete decision/regression tree */
  #endif                        /* and attribute set */
  #ifdef STORAGE
  showmem("at end of program"); /* check memory usage */
  #endif
  return 0;                     /* return 'ok' */
}  /* main() */