static void error (int code, ...)
{
#ifndef QUIET
va_list args;
const char *msg;
assert(prgname);
if (code < E_UNKNOWN) code = E_UNKNOWN;
if (code < 0) {
msg = errmsgs[-code];
if (!msg) msg = errmsgs[-E_UNKNOWN];
fprintf(stderr, "\n%s: ", prgname);
va_start(args, code);
vfprintf(stderr, msg, args);
va_end(args);
}
#endif
#ifndef NDEBUG
if (map) free(map);
if (isrep) isr_delete(isrep, 0);
if (istree) ist_delete(istree);
if (tatree) tt_delete(tatree, 0);
if (tabag) tb_delete(tabag, 0);
if (ibase) ib_delete(ibase);
if (in && (in != stdin)) fclose(in);
if (out && (out != stdout)) fclose(out);
#endif
#ifdef STORAGE
showmem("at end of program");
#endif
exit(code);
} /* error() */
static void error (int code, ...)
{ /* --- print an error message */
#ifndef QUIET /* if not quiet version */
va_list args; /* list of variable arguments */
const char *msg; /* error message */
assert(prgname); /* check the program name */
if (code < E_UNKNOWN) code = E_UNKNOWN;
if (code < 0) { /* if to report an error, */
msg = errmsgs[-code]; /* get the error message */
if (!msg) msg = errmsgs[-E_UNKNOWN];
fprintf(stderr, "\n%s: ", prgname);
va_start(args, code); /* get variable arguments */
vfprintf(stderr, msg, args);/* print error message */
va_end(args); /* end argument evaluation */
}
#endif
#ifndef NDEBUG /* if debug version */
if (istree) ist_delete(istree); /* clean up memory */
if (tatree) tat_delete(tatree); /* and close files */
if (taset) tas_delete(taset, 0);
if (itemset) is_delete(itemset);
if (in && (in != stdin)) fclose(in);
if (out && (out != stdout)) fclose(out);
#endif
#ifdef STORAGE /* if storage debugging */
showmem("at end of program"); /* check memory usage */
#endif
exit(code); /* abort the program */
} /* error() */
int do_apriori (int argc, char *argv[])
{ /* --- main function */
int i, k = 0, n; /* loop variables, counters */
char *s; /* to traverse the options */
char **optarg = NULL; /* option argument */
char *fn_in = NULL; /* name of input file */
char *fn_out = NULL; /* name of output file */
char *fn_app = NULL; /* name of item appearances file */
char *blanks = NULL; /* blanks */
char *fldseps = NULL; /* field separators */
char *recseps = NULL; /* record separators */
char *comment = NULL; /* comment indicators */
char *used = NULL; /* item usage vector */
double supp = 0.1; /* minimal support (in percent) */
double smax = 1.0; /* maximal support (in percent) */
double conf = 0.8; /* minimal confidence (in percent) */
int mode = IST_BODY; /* search mode (rule support def.) */
int target = 'r'; /* target type (sets/rules/h.edges) */
int arem = 0; /* additional rule evaluation measure */
int lift = 0; /* flag for printing the lift */
double minval = 0.1; /* minimal evaluation measure value */
double lftval = 0; /* lift value (confidence/prior) */
int minlen = 1; /* minimal rule length */
int maxlen = INT_MAX; /* maximal rule length */
int load = 1; /* flag for loading transactions */
int sort = 2; /* flag for item sorting and recoding */
double filter = 0.1; /* item usage filtering parameter */
int tree = 1; /* flag for transaction tree */
int heap = 1; /* flag for heap sort vs. quick sort */
int c2scf = 0; /* flag for conv. to scanable form */
char *sep = " "; /* item separator for output */
char *fmt = "%.1f"; /* output format for support/conf. */
int sout = 1; /* flag for abs./rel. support output */
int ext = 0; /* flag for extended support output */
int aval = 0; /* flag for add. eval. measure value */
int maxcnt = 0; /* maximal number of items per set */
int tacnt; /* number of transactions */
int frq; /* frequency of an item set */
int *map, *set; /* identifier map, item set */
int verbose = 0; /* flag for verboseness */
const char *name; /* buffer for item names */
static char buf[4*TS_SIZE+4]; /* buffer for formatting */
clock_t t, tt, tc, x; /* timer for measurements */
#ifndef QUIET /* if not quiet version */
prgname = argv[0]; /* get program name for error msgs. */
/* --- print 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 arguments given */
printf("usage: %s [options] infile outfile [appfile]\n", argv[0]);
printf("%s\n", DESCRIPTION);
printf("%s\n", VERSION);
printf("-t# target type (default: association rules)\n"
" (s: item sets, c: closed item sets,"
" m: maximal item sets,\n"
" r: association rules,"
" h: association hyperedges)\n");
printf("-m# minimal number of items per set/rule/hyperedge "
"(default: %d)\n", minlen);
printf("-n# maximal number of items per set/rule/hyperedge "
"(default: no limit)\n");
printf("-s# minimal support of a set/rule/hyperedge "
"(default: %g%%)\n", supp *100);
printf("-S# maximal support of a set/rule/hyperedge "
"(default: %g%%)\n", smax *100);
printf("-c# minimal confidence of a rule/hyperedge "
"(default: %g%%)\n", conf *100);
printf("-o use original definition of the support of a rule "
"(body & head)\n");
printf("-k# item separator for output "
"(default: \"%s\")\n", sep);
printf("-p# output format for support/confidence "
"(default: \"%s\")\n", fmt);
printf("-x extended support output "
"(print both rule support types)\n");
printf("-a print absolute support "
"(number of transactions)\n");
printf("-y print lift value (confidence divided by prior)\n");
printf("-e# additional evaluation measure (default: none)\n");
printf("-! print a list of additional evaluation measures\n");
printf("-d# minimal value of additional evaluation measure "
"(default: %g%%)\n", minval *100);
printf("-v print value of additional "
"rule evaluation measure\n");
printf("-g write output in scanable form "
"(quote certain characters)\n");
printf("-l do not load transactions into memory "
"(work on input file)\n");
printf("-q# sort items w.r.t. their frequency (default: %d)\n"
" (1: ascending, -1: descending, 0: do not sort,\n"
" 2: ascending, -2: descending w.r.t. "
"transaction size sum)\n", sort);
printf("-u# filter unused items from transactions "
"(default: %g)\n", filter);
printf(" (0: do not filter items w.r.t. usage in sets,\n"
" <0: fraction of removed items for filtering,\n"
" >0: take execution times ratio into account)\n");
printf("-h do not organize transactions as a prefix tree\n");
printf("-j use quicksort to sort the transactions "
"(default: heapsort)\n");
printf("-z minimize memory usage "
"(default: maximize speed)\n");
printf("-b/f/r# blank characters, field and record separators\n"
" (default: \" \\t\\r\", \" \\t\", \"\\n\")\n");
printf("-C# comment characters (default: \"#\")\n");
printf("-V verbose\n");
printf("infile file to read transactions from\n");
printf("outfile file to write item sets/association rules"
"/hyperedges to\n");
printf("appfile file stating item appearances (optional)\n");
return 0; /* print a usage message */
} /* and abort the program */
#endif /* #ifndef QUIET */
/* --- 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 (*s) { /* traverse options */
switch (*s++) { /* evaluate switches */
case '!': help(); break;
case 't': target = (*s) ? *s++ : 'r'; break;
case 'm': minlen = (int)strtol(s, &s, 0); break;
case 'n': maxlen = (int)strtol(s, &s, 0); break;
case 's': supp = 0.01*strtod(s, &s); break;
case 'S': smax = 0.01*strtod(s, &s); break;
case 'c': conf = 0.01*strtod(s, &s); break;
case 'o': mode |= IST_BOTH; break;
case 'k': optarg = &sep; break;
case 'p': optarg = &fmt; break;
case 'x': ext = 1; break;
case 'a': sout |= 2; break;
case 'y': lift = 1; break;
case 'e': arem = (*s) ? *s++ : 0; break;
case 'd': minval = 0.01*strtod(s, &s); break;
case 'v': aval = 1; break;
case 'g': c2scf = 1; break;
case 'l': load = 0; break;
case 'q': sort = (int)strtol(s, &s, 0); break;
case 'u': filter = strtod(s, &s); break;
case 'h': tree = 0; break;
case 'j': heap = 0; break;
case 'z': mode |= IST_MEMOPT; break;
case 'b': optarg = &blanks; break;
case 'f': optarg = &fldseps; break;
case 'r': optarg = &recseps; break;
case 'C': optarg = &comment; break;
case 'V': verbose = 1; break;
default : error(E_OPTION, *--s); break;
} /* set option variables */
if (optarg && *s) { *optarg = s; optarg = NULL; break; }
} } /* get option argument */
else { /* -- if argument is no option */
switch (k++) { /* evaluate non-options */
case 0: fn_in = s; break;
case 1: fn_out = s; break;
case 2: fn_app = s; break;
default: error(E_ARGCNT); break;
} /* note filenames */
}
}
if (optarg) error(E_OPTARG); /* check option argument */
if ((k < 2) || (k > 3)) /* and the number of arguments */
error(E_ARGCNT); /* (either in/out or in/out/app) */
if ((!fn_in || !*fn_in) && (fn_app && !*fn_app))
error(E_STDIN); /* stdin must not be used twice */
switch (target) { /* check and translate target type */
case 's': target = TT_SET; break;
case 'c': target = TT_CLSET; break;
case 'm': target = TT_MFSET; break;
case 'r': target = TT_RULE; break;
case 'h': target = TT_HEDGE; break;
case 'g': target = TT_GROUP; break;
default : error(E_TARGET, (char)target); break;
}
if (supp > 1) /* check the minimal support */
error(E_SUPP, supp); /* (< 0: absolute number) */
if ((conf < 0) || (conf > 1))
error(E_CONF, conf); /* check the minimal confidence */
if (minlen <= 0) error(E_RULELEN, minlen); /* check the limits */
if (maxlen <= 0) error(E_RULELEN, maxlen); /* for the rule length */
switch (arem) { /* check and translate measure */
case 0 : case '0': arem = EM_NONE; break;
case 'd': case '1': arem = EM_DIFF; break;
case 'q': case '2': arem = EM_QUOT; break;
case 'a': case '3': arem = EM_AIMP; break;
case 'i': case '4': arem = EM_INFO; break;
case 'c': case '5': arem = EM_CHI2; break;
case 'p': case '6': arem = EM_PVAL; break;
default : error(E_MEASURE, (char)arem); break;
}
if (target <= TT_MFSET) { /* in item set mode neutralize */
mode |= IST_BOTH; conf = 1;}/* rule specific settings */
if (arem == EM_NONE) /* if no add. rule eval. measure, */
aval = 0; /* clear the corresp. output flag */
if ((filter <= -1) || (filter >= 1)) filter = 0;
/* --- create item set and transaction set --- */
itemset = is_create(-1); /* create an item set and */
if (!itemset) error(E_NOMEM); /* set the special characters */
is_chars(itemset, blanks, fldseps, recseps, comment);
if (load) { /* if to load the transactions */
taset = tas_create(itemset);
if (!taset) error(E_NOMEM); /* create a transaction set */
} /* to store the transactions */
MSG(fprintf(stderr, "\n")); /* terminate the startup message */
/* --- read item appearances --- */
if (fn_app) { /* if item appearances are given */
t = clock(); /* start the timer */
if (*fn_app) /* if an app. file name is given, */
in = fopen(fn_app, "r"); /* open the item appearances file */
else { /* if no app. file name is given, */
in = stdin; fn_app = "<stdin>"; } /* read from std. input */
MSG(fprintf(stderr, "reading %s ... ", fn_app));
if (!in) error(E_FOPEN, fn_app);
k = is_readapp(itemset,in); /* read the item appearances */
if (k != 0) error(k, fn_app, RECCNT(itemset), BUFFER(itemset));
if (in != stdin) /* if not read from standard input, */
fclose(in); /* close the input file */
MSG(fprintf(stderr, "[%d item(s)]", is_cnt(itemset)));
MSG(fprintf(stderr, " done [%.2fs].\n", SEC_SINCE(t)));
} /* print a log message */
/* --- read transactions --- */
t = clock(); /* start the timer */
if (fn_in && *fn_in) /* if an input file name is given, */
in = fopen(fn_in, "r"); /* open input file for reading */
else { /* if no input file name is given, */
in = stdin; fn_in = "<stdin>"; } /* read from standard input */
MSG(fprintf(stderr, "reading %s ... \n", fn_in));
if (!in) error(E_FOPEN, fn_in);
while (1) { /* transaction read loop */
k = is_read(itemset, in); /* read the next transaction */
if (k < 0) error(k, fn_in, RECCNT(itemset), BUFFER(itemset));
if (k > 0) break; /* check for error and end of file */
k = is_tsize(itemset); /* update the maximal */
if (k > maxcnt) maxcnt = k; /* transaction size */
if (taset && (tas_add(taset, NULL, 0) != 0))
error(E_NOMEM); /* add the loaded transaction */
} /* to the transaction set */
if (taset) { /* if transactions have been loaded */
if (in != stdin) fclose(in);/* if not read from standard input, */
in = NULL; /* close the input file */
} /* clear the file variable */
n = is_cnt(itemset); /* get the number of items */
tacnt = is_gettac(itemset); /* and the number of transactions */
MSG(fprintf(stderr, "[%d item(s), %d transaction(s)]", n, tacnt));
MSG(fprintf(stderr, " done [%.2fs].", SEC_SINCE(t)));
if ((n <= 0) || (tacnt <= 0)) error(E_NOTAS);
MSG(fprintf(stderr, "\n")); /* check for at least one transaction */
if (supp >= 0) /* if relative support is given */
supp = ceil(tacnt *supp); /* compute absolute support */
else { /* if absolute support is given, */
supp = ceil(-100 *supp); /* make the support value positive */
if (!(sout & 2)) sout = 2; /* switch to absolute support output */
} /* do the same with the max. support */
smax = floor(((smax >= 0) ? tacnt : -100) *smax);
/* --- sort and recode items --- */
MSG(fprintf(stderr, "filtering, sorting and recoding items ... "));
t = clock(); /* start the timer */
map = (int*)malloc(is_cnt(itemset) *sizeof(int));
if (!map) error(E_NOMEM); /* create an item identifier map */
k = (int)((mode & IST_HEAD) ? supp : ceil(supp *conf));
n = is_recode(itemset, k, sort, map);
if (taset) { /* sort and recode the items and */
tas_recode(taset, map,n); /* recode the loaded transactions */
maxcnt = tas_max(taset); /* get the new maximal t.a. size */
} /* (may be smaller than before) */
free(map); /* delete the item identifier map */
MSG(fprintf(stderr, "[%d item(s)] ", n));
MSG(fprintf(stderr, "done [%.2fs].", SEC_SINCE(t)));
if (n <= 0) error(E_NOFREQ); /* print a log message and */
MSG(fprintf(stderr, "\n")); /* check the number of items */
if (maxlen > maxcnt) /* clamp the set/rule length */
maxlen = maxcnt; /* to the maximum set size */
/* --- create a transaction tree --- */
tt = 0; /* init. the tree construction time */
if (tree && taset) { /* if transactions were loaded */
MSG(fprintf(stderr, "creating transaction tree ... "));
t = clock(); /* start the timer */
tatree = tat_create(taset, heap);
if (!tatree) error(E_NOMEM);/* create a transaction tree */
if (filter == 0) { /* if a tree rebuild is not needed, */
tas_delete(taset, 0); taset = NULL; } /* delete transactions */
tt = clock() -t; /* note the time for the construction */
MSG(fprintf(stderr, "done [%.2fs].\n", SEC_SINCE(t)));
} /* print a log message */
/* --- create an item set tree --- */
t = clock(); tc = 0; /* start the timer */
istree = ist_create(itemset, mode, (int)supp, conf);
if (!istree) error(E_NOMEM); /* create an item set tree */
/* --- check item subsets --- */
if (filter) { /* if to filter unused items */
used = (char*)malloc(is_cnt(itemset) *sizeof(char));
if (!used) error(E_NOMEM); /* create a flag vector */
} /* for the items */
MSG(fprintf(stderr, "checking subsets of size 1"));
while (ist_height(istree) < maxlen) {
if (filter != 0) { /* if to filter w.r.t. item usage, */
i = ist_check(istree, used); /* check current item usage */
if (i < maxlen) maxlen = i; /* update the maximum size */
if (ist_height(istree) >= i) break;
} /* check the tree height */
k = ist_addlvl(istree); /* while max. height is not reached, */
if (k < 0) error(E_NOMEM); /* add a level to the item set tree */
if (k != 0) break; /* if no level was added, abort */
MSG(fprintf(stderr, " %d", ist_height(istree)));
if (tatree) { /* if a transaction tree was created */
if (((filter < 0) /* if to filter w.r.t. item usage */
&& (i < -filter *n)) /* and enough items were removed */
|| ((filter > 0) /* or counting time is long enough */
&& (i < n) && (i *(double)tt < filter *n *tc))) {
n = i; x = clock(); /* note the new number of items */
tas_filter(taset, used);/* and remove unnecessary items */
tat_delete(tatree); /* delete the transaction tree */
tatree = tat_create(taset, heap);
if (!tatree) error(E_NOMEM);
tt = clock() -x; /* rebuild the transaction tree and */
} /* note the new construction time */
x = clock(); /* count the transaction tree */
ist_countx(istree, tatree);
tc = clock() -x; } /* note the new count time */
else if (taset) { /* if transactions were loaded */
if (((filter < 0) /* if to filter w.r.t. item usage */
&& (i <= -filter *n)) /* and enough items were removed */
|| ((filter > 0) /* or counting time is long enough */
&& (i *(double)tt <= filter *n *tc))) {
n = i; x = clock(); /* note the new number of items */
tas_filter(taset, used);/* and remove unnecessary items */
tt = clock() -t; /* from the transactions */
} /* note the filtering time */
for (i = tacnt; --i >= 0;)/* traverse and count transactions */
ist_count(istree, tas_tract(taset, i), tas_tsize(taset, i));
tc = clock() -t; } /* note the new count time */
else { /* if to work on the input file, */
rewind(in); /* reset the file position */
for (maxcnt = 0; (i = is_read(itemset, in)) == 0; ) {
if (filter != 0) /* (re)read the transactions and */
is_filter(itemset, used); /* remove unnecessary items */
k = is_tsize(itemset); /* update the maximum size */
if (k > maxcnt) maxcnt = k; /* of a transaction */
ist_count(istree, is_tract(itemset), k);
} /* count the transaction in the tree */
if (i < 0) error(i, fn_in, RECCNT(itemset), BUFFER(itemset));
if (maxcnt < maxlen) /* update the maximal rule length */
maxlen = maxcnt; /* according to the max. t.a. size */
} /* (may be smaller than before) */
}
if (!taset && !tatree) { /* if transactions were not loaded */
if (in != stdin) fclose(in);/* if not read from standard input, */
in = NULL; /* close the input file */
} /* clear the file variable */
MSG(fprintf(stderr, " done [%.2fs].\n", SEC_SINCE(t)));
/* --- filter found item sets --- */
if ((target == TT_CLSET) || (target == TT_MFSET)) {
MSG(fprintf(stderr, "filtering %s item sets ... ",
(target == TT_MFSET) ? "maximal" : "closed"));
t = clock(); /* filter the item sets */
ist_filter(istree, (target == TT_MFSET) ? IST_MAXFRQ : IST_CLOSED);
MSG(fprintf(stderr, "done [%.2fs].\n", SEC_SINCE(t)));
} /* (filter takes longer than print) */
/* --- sort transactions --- */
if (target <= TT_MFSET) { /* if to find frequent item sets */
if (!taset) /* transactions must be loaded */
ext = 0; /* for extended support output */
else if (ext) { /* if extended output is requested */
MSG(fprintf(stderr, "sorting transactions ... "));
t = clock(); /* start the timer */
tas_sort(taset, heap); /* sort the transactions */
MSG(fprintf(stderr, "done [%.2fs].\n", SEC_SINCE(t)));
} /* (sorting is necessary to find the */
} /* number of identical transactions) */
/* --- print item sets/rules/hyperedges --- */
t = clock(); /* start the timer */
if (fn_out && *fn_out) /* if an output file name is given, */
out = fopen(fn_out, "w"); /* open the output file */
else { /* if no output file name is given, */
out = stdout; fn_out = "<stdout>"; } /* write to std. output */
MSG(fprintf(stderr, "writing %s ... ", fn_out));
if (!out) error(E_FOPEN, fn_out);
ist_init(istree, minlen, arem, minval);
set = is_tract(itemset); /* get the transaction buffer */
if (target <= TT_MFSET) { /* if to find frequent item sets */
for (n = 0; 1; ) { /* extract item sets from the tree */
k = ist_set(istree, set, &frq, &conf);
if (k <= 0) break; /* get the next frequent item set */
if (frq > smax) continue; /* check against maximal support */
for (i = 0; i < k; i++) { /* traverse the set's items */
name = is_name(itemset, set[i]);
if (c2scf) { sc_format(buf, name, 0); name = buf; }
fputs(name, out); /* print the name of the next item */
fputs((i < k-1) ? sep : " ", out);
} /* print a separator */
fputs(" (", out); /* print the item set's support */
if (sout & 1) { fprintf(out, fmt, (frq/(double)tacnt) *100);
if (sout & 2) fputc('/', out); }
if (sout & 2) { fprintf(out, "%d", frq); }
if (ext) { /* if to print the extended support */
frq = tas_occur(taset, set, k);
fputs(", ", out); /* get the number of occurrences */
fprintf(out, fmt, (frq/(double)tacnt) *100);
if (sout & 2) fprintf(out, "/%d", frq);
} /* print the extended support data */
if (aval) { fputs(", ", out); fprintf(out, fmt, conf *100); }
fputs(")\n", out); /* print the add. eval. measure, */
n++; /* terminate the support output, */
} } /* and count the item set */
else if (target == TT_RULE) { /* if to find association rules, */
for (n = 0; 1; ) { /* extract rules from tree */
k = ist_rule(istree, set, &frq, &conf, &lftval, &minval);
if (k <= 0) break; /* get the next association rule */
if (frq > smax) continue; /* check against maximal support */
for (i = 0; i < k; i++) { /* traverse the rule's items */
name = is_name(itemset, set[i]);
if (c2scf) { sc_format(buf, name, 0); name = buf; }
fputs(name, out); /* print the next item */
fputs((i <= 0) ? " <- " : ((i < k-1) ? sep : " "), out);
} /* print a separator */
fputs(" (", out); /* print the rule evaluation */
if (sout & 1) supp = frq/(double)tacnt;
if (ext && !(mode & IST_HEAD)) {
if (sout & 1) { fprintf(out, fmt, supp *conf *100);
if (sout & 2) fputc('/', out); }
if (sout & 2) { fprintf(out, "%d", (int)(frq *conf +0.5));}
fputs(", ", out); /* print the support of the rule */
} /* from the support of the body */
if (sout & 1) { fprintf(out, fmt, supp *100);
if (sout & 2) fputc('/', out); }
if (sout & 2) { fprintf(out, "%d", frq); }
fputs(", ", out); /* print the rule support */
if (ext && (mode & IST_HEAD)) {
if (sout & 1) { fprintf(out, fmt, (supp/conf) *100);
if (sout & 2) fputc('/', out); }
if (sout & 2) { fprintf(out, "%d", (int)(frq /conf +0.5));}
fputs(", ", out); /* print the support of the body */
} /* from the support of the rule */
fprintf(out, fmt, conf *100); /* print the rule confidence */
if (lift) { fputs(", ", out); fprintf(out, fmt, lftval *100); }
if (aval) { fputs(", ", out); fprintf(out, fmt, minval *100); }
fputs(")\n", out); /* print the value of the additional */
n++; /* rule evaluation measure and */
} } /* count the association rule */
else if (target == TT_HEDGE){ /* if to find association hyperedges */
for (n = 0; 1; ) { /* extract hyperedges from tree */
k = ist_hedge(istree, set, &frq, &conf, &minval);
if (k <= 0) break; /* get the next hyperedge */
if (frq > smax) continue; /* check against maximal support */
for (i = 0; i < k; i++) { /* traverse the edge's items */
name = is_name(itemset, set[i]);
if (c2scf) { sc_format(buf, name, 0); name = buf; }
fputs(name, out); /* print the name of the next item */
fputs((i < k-1) ? sep : " ", out);
} /* print a separator */
fputs(" (", out); /* print the hyperedge evaluation */
if (sout & 1) { fprintf(out, fmt, (frq/(double)tacnt) *100);
if (sout & 2) fputc('/', out); }
if (sout & 2) { fprintf(out, "%d", frq); }
fputs(", ", out); fprintf(out, fmt, conf *100);
if (aval) { fputs(", ", out); fprintf(out, fmt, minval *100); }
fputs(")\n", out); /* print support and confidence */
n++; /* of the hyperedge and */
} } /* count the hyperedge */
else { /* if to find association groups */
for (n = 0; 1; ) { /* extract groups from tree */
k = ist_group(istree, set, &frq, &minval);
if (k <= 0) break; /* get the next group */
if (frq > smax) continue; /* check against maximal support */
for (i = 0; i < k; i++) { /* traverse the group's items */
name = is_name(itemset, set[i]);
if (c2scf) { sc_format(buf, name, 0); name = buf; }
fputs(name, out); /* print the name of the next item */
fputs((i < k-1) ? sep : " ", out);
} /* print a separator */
fputs(" (", out); /* print the group evaluation */
if (sout & 1) { fprintf(out, fmt, (frq/(double)tacnt) *100);
if (sout & 2) fputc('/', out); }
if (sout & 2) { fprintf(out, "%d", frq); }
if (aval) { fputs(", ", out); fprintf(out, fmt, minval *100); }
fputs(")\n", out); /* print support and add. measure */
n++; /* and count the group */
}
} /* if (target <= TT_MFSET) .. else .. */
if (fflush(out) != 0) error(E_FWRITE, fn_out);
if (out != stdout) fclose(out);
out = NULL; /* close the output file */
MSG(fprintf(stderr, "[%d %s(s)] done ", n, ttypes[target]));
MSG(fprintf(stderr, "[%.2fs].\n", SEC_SINCE(t)));
#ifdef BENCH
printf("number of support counters: %d\n", istree->sccnt);
printf("necessary support counters: %d\n", istree->scnec);
printf("number of child pointers : %d\n", istree->cpcnt);
printf("necessary child pointers : %d\n", istree->cpnec);
printf("allocated memory (bytes) : %d\n", istree->bytes);
#endif
/* --- clean up --- */
#ifndef NDEBUG /* if this is a debug version */
free(used); /* delete the item app. vector */
ist_delete(istree); /* delete the item set tree, */
if (tatree) tat_delete(tatree); /* the transaction tree, */
if (taset) tas_delete(taset, 0); /* the transaction set, */
is_delete(itemset); /* and the item set */
#endif
#ifdef STORAGE /* if storage debugging */
showmem("at end of program"); /* check memory usage */
#endif
return 0; /* return 'ok' */
} /* main() */
int main (int argc, char *argv[])
{ /* --- main function */
int i, k = 0, n; /* loop variables, counters */
char *s; /* to traverse the options */
char **optarg = NULL; /* option argument */
char *fn_in = NULL; /* name of input file */
char *fn_out = NULL; /* name of output file */
char *fn_app = NULL; /* name of item appearances file */
char *blanks = NULL; /* blanks */
char *fldseps = NULL; /* field separators */
char *recseps = NULL; /* record separators */
char *comment = NULL; /* comment indicators */
char *isep = " "; /* item separator for output */
char *impl = " <- "; /* implication sign for ass. rules */
char *dflt = " (%1S)"; /* default format for check */
char *format = dflt; /* format for information output */
int target = 's'; /* target type (sets/rules/h.edges) */
int min = 1; /* minimum rule/item set size */
int max = INT_MAX; /* maximum rule/item set size */
double supp = 10; /* minimum support (in percent) */
double smax = 100; /* maximum support (in percent) */
double conf = 80; /* minimum confidence (in percent) */
int dir = 0; /* direction for size sorting */
int eval = 0; /* additional evaluation measure */
int zero = 0; /* flag for zero eval. below expect. */
int aggm = 0; /* aggregation mode for eval. measure */
double minval = 10; /* minimum evaluation measure value */
int prune = 0; /* (min. size for) evaluation pruning */
double filter = 0.1; /* item usage filtering parameter */
int sort = 2; /* flag for item sorting and recoding */
int tree = 1; /* flag for transaction tree */
int heap = 1; /* flag for heap sort vs. quick sort */
int post = 0; /* flag for a-posteriori pruning */
int report = 0; /* other flags for reporting */
int mode = APP_BODY|IST_PERFECT; /* search mode */
int size; /* current item set size */
int wgt; /* total transaction weight */
int frq, body, head; /* frequency of an item set */
int *items; /* item set (for reporting) */
clock_t t, tt, tc, x; /* timers for measurements */
#ifndef QUIET /* if not quiet version */
prgname = argv[0]; /* get program name for error msgs. */
/* --- print 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 arguments are given */
printf("usage: %s [options] infile outfile [appfile]\n", argv[0]);
printf("%s\n", DESCRIPTION);
printf("%s\n", VERSION);
printf("-t# target type "
"(default: %c)\n", target);
printf(" (s: frequent item sets, c: closed item sets,\n"
" m: maximal item sets, r: association rules)\n");
printf("-m# minimum number of items per set/rule "
"(default: %d)\n", min);
printf("-n# maximum number of items per set/rule "
"(default: no limit)\n");
printf("-s# minimum support of a set/rule "
"(default: %g%%)\n", supp);
printf("-S# maximum support of a set/rule "
"(default: %g%%)\n", smax);
printf(" (positive: percentage, "
"negative: absolute number)\n");
printf("-c# minimum confidence of a rule "
"(default: %g%%)\n", conf);
printf("-o use the original rule support definition "
"(body & head)\n");
printf("-e# additional evaluation measure "
"(default: none)\n");
printf("-a# aggregation mode for evaluation measure "
"(default: none)\n");
printf("-z zero evaluation below expected support "
"(default: evaluate all)\n");
printf("-d# minimum value of add. evaluation measure "
"(default: %g%%)\n", minval);
printf("-p# (min. size for) pruning with evaluation "
"(default: no pruning)\n");
printf(" (< 0: backward, > 0: forward)\n");
printf("-l# sort item sets in output by their size "
"(default: no sorting)\n");
printf(" (< 0: descending, > 0: ascending)\n");
printf("-g write item names in scanable form "
"(quote certain characters)\n");
printf("-k# item separator for output "
"(default: \"%s\")\n", isep);
printf("-i# implication sign for association rules "
"(default: \"%s\")\n", impl);
printf("-v# output format for set/rule information "
"(default: \"%s\")\n", format);
printf("-q# sort items w.r.t. their frequency "
"(default: %d)\n", sort);
printf(" (1: ascending, -1: descending, 0: do not sort,\n"
" 2: ascending, -2: descending w.r.t. "
"transaction size sum)\n");
printf("-u# filter unused items from transactions "
"(default: %g)\n", filter);
printf(" (0: do not filter items w.r.t. usage in sets,\n"
" <0: fraction of removed items for filtering,\n"
" >0: take execution times ratio into account)\n");
printf("-j use quicksort to sort the transactions "
"(default: heapsort)\n");
printf("-x do not prune the search "
"with perfect extensions\n");
printf("-y a-posteriori pruning of infrequent item sets\n");
printf("-h do not organize transactions as a prefix tree\n");
printf("-b# blank characters "
"(default: \" \\t\\r\")\n");
printf("-f# field separators "
"(default: \" \\t,\")\n");
printf("-r# record separators "
"(default: \"\\n\")\n");
printf("-C# comment characters "
"(default: \"#\")\n");
printf("-! print additional option information\n");
printf("infile file to read transactions from\n");
printf("outfile file to write item sets/association rules"
"/hyperedges to\n");
printf("appfile file stating item appearances (optional)\n");
return 0; /* print a usage message */
} /* and abort the program */
#endif /* #ifndef QUIET */
/* free option characters: w [A-Z]\[SC] */
/* --- evaluate arguments --- */
for (i = 1; i < argc; i++) { /* traverse the arguments */
s = argv[i]; /* get an option argument */
if (optarg) { *optarg = s; optarg = NULL; continue; }
if ((*s == '-') && *++s) { /* -- if argument is an option */
while (*s) { /* traverse the options */
switch (*s++) { /* evaluate the options */
case '!': help(); break;
case 't': target = (*s) ? *s++ : 's'; break;
case 'm': min = (int)strtol(s, &s, 0); break;
case 'n': max = (int)strtol(s, &s, 0); break;
case 's': supp = strtod(s, &s); break;
case 'S': smax = strtod(s, &s); break;
case 'c': conf = strtod(s, &s); break;
case 'o': mode |= APP_BOTH; break;
case 'e': eval = (*s) ? *s++ : 0; break;
case 'z': zero = IST_ZERO; break;
case 'a': aggm = (*s) ? *s++ : 0; break;
case 'd': minval = strtod(s, &s); break;
case 'p': prune = (int)strtol(s, &s, 0); break;
case 'g': report = ISR_SCAN; break;
case 'k': optarg = &isep; break;
case 'i': optarg = &impl; break;
case 'v': optarg = &format; break;
case 'l': dir = (int)strtol(s, &s, 0); break;
case 'q': sort = (int)strtol(s, &s, 0); break;
case 'u': filter = strtod(s, &s); break;
case 'h': tree = 0; break;
case 'j': heap = 0; break;
case 'x': mode &= ~IST_PERFECT; break;
case 'y': post = 1; break;
case 'b': optarg = &blanks; break;
case 'f': optarg = &fldseps; break;
case 'r': optarg = &recseps; break;
case 'C': optarg = &comment; break;
default : error(E_OPTION, *--s); break;
} /* set the option variables */
if (optarg && *s) { *optarg = s; optarg = NULL; break; }
} } /* get an option argument */
else { /* -- if argument is no option */
switch (k++) { /* evaluate non-options */
case 0: fn_in = s; break;
case 1: fn_out = s; break;
case 2: fn_app = s; break;
default: error(E_ARGCNT); break;
} /* note filenames */
}
}
if (optarg) error(E_OPTARG); /* check the option argument */
if ((k < 2) || (k > 3)) /* and the number of arguments */
error(E_ARGCNT); /* (either in/out or in/out/app) */
if ((!fn_in || !*fn_in) && (fn_app && !*fn_app))
error(E_STDIN); /* stdin must not be used twice */
switch (target) { /* check and translate target type */
case 's': target = TT_ALL; break;
case 'c': target = TT_CLOSED; break;
case 'm': target = TT_MAXIMAL; break;
case 'r': target = TT_RULE; break;
default : error(E_TARGET, (char)target); break;
}
if (min < 0) error(E_SIZE, min); /* check the limits */
if (max < 0) error(E_SIZE, max); /* for the set size */
if (supp > 100) /* check the minimum support */
error(E_SUPP, supp); /* (< 0: absolute support) */
if ((conf < 0) || (conf > 100))
error(E_CONF, conf); /* check the minimum confidence */
switch (eval) { /* check and translate measure */
case 'x': case 0: eval = IST_NONE; break;
case 'c': eval = IST_CONF; break;
case 'd': eval = IST_CONF_DIFF; break;
case 'l': eval = IST_LIFT; break;
case 'a': eval = IST_LIFT_DIFF; break;
case 'q': eval = IST_LIFT_QUOT; break;
case 'v': eval = IST_CVCT; break;
case 'e': eval = IST_CVCT_DIFF; break;
case 'r': eval = IST_CVCT_QUOT; break;
case 'f': eval = IST_CERT; break;
case 'n': eval = IST_CHI2; break;
case 'p': eval = IST_CHI2_PVAL; break;
case 'i': eval = IST_INFO; break;
case 'g': eval = IST_INFO_PVAL; break;
case 'b': eval = IST_LOGQ; break;
default : error(E_MEASURE, (char)eval); break;
}
switch (aggm) { /* check and translate agg. mode */
case 'x': case 0: aggm = IST_NONE; break;
case 'm': aggm = IST_MIN; break;
case 'n': aggm = IST_MAX; break;
case 'a': aggm = IST_AVG; break;
default : error(E_MEASURE, (char)aggm); break;
}
if ((target > TT_ALL) /* if individual set counters needed */
|| ((eval > IST_NONE) && (eval < IST_LOGQ)))
mode &= ~IST_PERFECT; /* remove perfect extension pruning */
if (target <= TT_MAXIMAL) { /* remove rule specific settings */
mode |= APP_BOTH; conf = 100; }
if ((filter <= -1) || (filter >= 1))
filter = 0; /* check and adapt the filter option */
if (dir) /* if to sort output by size, */
mode &= ~IST_PERFECT; /* do not use perfect ext. pruning */
/* --- create item base --- */
ibase = ib_create(0, 0); /* create an item base and */
if (!ibase) error(E_NOMEM); /* set the special characters */
ib_chars(ibase, blanks, fldseps, recseps, "", comment);
MSG(stderr, "\n"); /* terminate the startup message */
/* --- read item appearance indicators --- */
if (fn_app) { /* if item appearances are given */
t = clock(); /* start the timer for the reading */
if (*fn_app) /* if an app. file name is given, */
in = fopen(fn_app, "r"); /* open the item appearances file */
else { /* if no app. file name is given, */
in = stdin; fn_app = "<stdin>"; } /* read from std. input */
MSG(stderr, "reading %s ... ", fn_app);
if (!in) error(E_FOPEN, fn_app);
k = ib_readapp(ibase, in); /* read the item appearances */
if (k != 0) error(k, fn_app, RECCNT(ibase), BUFFER(ibase));
if (in != stdin) fclose(in);/* if not read from standard input, */
in = NULL; /* close the input file */
MSG(stderr, "[%d item(s)]", ib_cnt(ibase));
MSG(stderr, " done [%.2fs].\n", SEC_SINCE(t));
} /* print a log message */
/* --- read transactions --- */
t = clock(); /* start the timer for the reading */
if (fn_in && *fn_in) /* if an input file name is given, */
in = fopen(fn_in, "r"); /* open input file for reading */
else { /* if no input file name is given, */
in = stdin; fn_in = "<stdin>"; } /* read from standard input */
MSG(stderr, "reading %s ... ", fn_in);
if (!in) error(E_FOPEN, fn_in);
tabag = tb_create(ibase, 0); /* create a transaction bag/multiset */
if (!tabag) error(E_NOMEM); /* to store the transactions */
while (1) { /* transaction read loop */
k = ib_read(ibase, in); /* read the next transaction */
if (k) { if (k > 0) break; /* check for error and end of file */
error(k, fn_in, RECCNT(ibase), BUFFER(ibase)); }
if (tb_add(tabag, NULL) != 0) error(E_NOMEM);
} /* add transaction to bag/multiset */
if (in != stdin) fclose(in); /* if not read from standard input, */
in = NULL; /* close the input file */
n = ib_cnt(ibase); /* get the number of items */
k = tb_cnt(tabag); /* get the number of transactions */
wgt = tb_wgt(tabag); /* the total transaction weight */
MSG(stderr, "[%d item(s), ", n);
if (k == wgt) MSG(stderr, "%d transaction(s)]", k);
else MSG(stderr, "%d/%d transaction(s)]", k, wgt);
MSG(stderr, " done [%.2fs].", SEC_SINCE(t));
if ((n <= 0) || (wgt <= 0)) /* check for at least one item */
error(E_NOTRANS); /* and at least one transaction */
MSG(stderr, "\n"); /* terminate the log message */
if (format == dflt) { /* if default info. format is used */
if (target != TT_RULE) format = (supp < 0) ? " (%a)" : " (%1S)";
else format = (supp < 0) ? " (%b, %1C)" : " (%1X, %1C)";
} /* set default according to target */
supp = ceil ((supp >= 0) ? 0.01 *supp *wgt : -supp);
smax = floor((smax >= 0) ? 0.01 *smax *wgt : -smax);
conf *= 0.01; /* transform support and confidence */
/* --- sort and recode items --- */
t = clock(); /* compute absolute support values */
MSG(stderr, "filtering, sorting and recoding items ... ");
map = (int*)malloc(n *sizeof(int));
if (!map) error(E_NOMEM); /* create an item identifier map */
k = (int)((mode & APP_HEAD) ? supp : ceil(supp *conf));
n = ib_recode(ibase, k, sort, map);
tb_recode(tabag, map); /* recode the items and transactions */
tb_itsort(tabag, 1, heap); /* and sort items in transactions */
free(map); map = NULL; /* delete the item identifier map */
MSG(stderr, "[%d item(s)] done [%.2fs].", n, SEC_SINCE(t));
if (n <= 0) error(E_NOFREQ); /* print a log message and */
MSG(stderr, "\n"); /* check the number of items */
k = tb_max(tabag); /* clamp the set/rule length to */
if (max > k) max = k; /* the maximum transaction size */
/* --- reduce transactions --- */
t = clock(); /* start the timer for the reduction */
MSG(stderr, "reducing transactions ... ");
tb_filter(tabag, min, NULL); /* remove items of short transactions */
tb_sort(tabag, 1, heap); /* sort the trans. lexicographically */
k = tb_reduce(tabag); /* reduce transactions to unique ones */
if (k == wgt) MSG(stderr, "[%d transaction(s)]", k);
else MSG(stderr, "[%d/%d transaction(s)]", k, wgt);
MSG(stderr, " done [%.2fs].\n", SEC_SINCE(t));
/* --- create transaction tree --- */
tt = 0; /* init. the tree construction time */
if (tree) { /* if to use a transaction tree */
t = clock(); /* start the timer for construction */
MSG(stderr, "building transaction tree ... ");
tatree = tt_create(tabag); /* create a transaction tree */
if (!tatree) error(E_NOMEM);
if (filter == 0) { /* if not to filter items, */
tb_delete(tabag, 0); /* delete the transaction bag */
tabag = NULL; /* (redundant data storage) */
}
MSG(stderr, "[%d node(s)]", tt_nodecnt(tatree));
MSG(stderr, " done [%.2fs].\n", SEC_SINCE(t));
tt = clock() -t; /* note the time for the construction */
} /* of the transaction tree */
/* --- create item set tree --- */
t = clock(); tc = 0; /* start the timer for the search */
istree = ist_create(ibase, mode, (int)supp, (int)smax, conf);
if (!istree) error(E_NOMEM); /* create an item set tree */
ist_seteval(istree, eval|zero, aggm, 0.01*minval, prune);
/* --- check item subsets --- */
MSG(stderr, "checking subsets of size 1");
map = (int*)malloc(n *sizeof(int));
if (!map) error(E_NOMEM); /* create a filter map */
while (1) { /* traverse the item set sizes */
size = ist_height(istree); /* get the current item set size and */
if (size >= max) break; /* abort if maximal size is reached */
if ((filter != 0) /* if to filter w.r.t. item usage */
&& (ist_check(istree, map) <= size))
break; /* check which items are still used */
if (post) /* if a-posteriori pruning requested, */
ist_prune(istree); /* prune infrequent item sets */
k = ist_addlvl(istree); /* while max. height is not reached, */
if (k) { if (k > 0) break; /* add a level to the item set tree */
error(E_NOMEM); } /* if no level was added, abort */
if (((filter < 0) /* if to filter w.r.t. item usage */
&& (i < -filter *n)) /* and enough items were removed */
|| ((filter > 0) /* or counting time is long enough */
&& (i < n) && (i *(double)tt < filter *n *tc))) {
n = i; /* note the new number of items */
x = clock(); /* start the timer for filtering */
tb_filter(tabag, size+1, map);
tb_sort(tabag, 0, heap); /* remove unnec. items and trans. */
tb_reduce(tabag); /* and reduce trans. to unique ones */
if (tatree) { /* if a transaction tree was created */
tt_delete(tatree, 0); /* delete the transaction tree */
tatree = tt_create(tabag);
if (!tatree) error(E_NOMEM);
} /* rebuild the transaction tree */
tt = clock() -x; /* note the filter/rebuild time */
}
MSG(stderr, " %d", ++size); /* print the current item set size */
x = clock(); /* start the timer for counting */
if (tatree) ist_countx(istree, tatree);
else ist_countb(istree, tabag);
tc = clock() -x; /* count the transaction tree/bag */
} /* and compute the new counting time */
free(map); map = NULL; /* delete the filter map */
MSG(stderr, " done [%.2fs].\n", SEC_SINCE(t));
/* --- filter found item sets --- */
if ((target == TT_CLOSED) || (target == TT_MAXIMAL)) {
t = clock(); /* start the timer for filtering */
MSG(stderr, "filtering for %s item sets ... ",
(target == TT_MAXIMAL) ? "maximal" : "closed");
k = target | ((prune < 0) ? IST_EVAL : 0);
ist_mark(istree, k); /* filter closed/maximal item sets */
MSG(stderr, "done [%.2fs].\n", SEC_SINCE(t));
} /* print a log message */
/* --- print item sets/rules/hyperedges --- */
t = clock(); /* start the timer for the output */
if (fn_out && *fn_out) /* if an output file name is given, */
out = fopen(fn_out, "w"); /* open the output file */
else { /* if no output file name is given, */
out = stdout; fn_out = "<stdout>"; } /* write to std. output */
MSG(stderr, "writing %s ... ", fn_out);
if (!out) error(E_FOPEN, fn_out);
if (eval == IST_LOGQ) report |= ISR_LOGS;
if ((target == TT_CLOSED) || (target == TT_MAXIMAL))
report |= ISR_NOEXP; /* combine the report mode flags */
isrep = isr_create(ibase, out, report, isep, impl);
if (!isrep) error(E_NOMEM); /* create an item set reporter */
isr_setfmt (isrep, format); /* and configure it */
isr_setsize(isrep, min, max);
ist_setsize(istree, min, max, dir);
ist_init (istree); /* initialize the extraction */
items = t_items(ib_tract(ibase));
if ((target <= TT_MAXIMAL) /* if to find frequent item sets */
&& (dir == 0)) { /* and not to sort them by size */
if (eval == IST_LOGQ) /* if to compute an add. evaluation */
isr_seteval(isrep, isr_logq, NULL, 0.01*minval);
else if (eval > IST_NONE) /* set the add. evaluation function */
isr_seteval(isrep, ist_evalx, istree, 0.01*minval);
n = ist_report(istree, isrep); } /* report the item sets */
else if (target <= TT_MAXIMAL) { /* if to find frequent item sets */
for (n = 0; 1; ) { /* extract item sets from the tree */
k = ist_set(istree, items, &frq, &minval);
if (k < 0) break; /* get the next frequent item set */
if (k > 0) fputs(isr_name(isrep, items[0]), out);
for (i = 0; ++i < k; ) { /* print the item names */
fputs(isep, out); fputs(isr_name(isrep, items[i]), out); }
if (format) /* if requested, print information */
isr_sinfo(isrep, frq, minval);
fputc('\n', out); n++; /* terminate the output line and */
} } /* count the reported item set */
else if (target == TT_RULE) { /* if to find association rules, */
for (n = 0; 1; ) { /* extract rules from tree */
k = ist_rule(istree, items, &frq, &body, &head, &minval);
if (k < 0) break; /* get the next association rule */
fputs(isr_name(isrep, items[0]), out);
fputs(impl, out); /* print name of rule head item */
if (k > 1) fputs(isr_name(isrep, items[1]), out);
for (i = 1; ++i < k; ) { /* print names of items in rule body */
fputs(isep, out); fputs(isr_name(isrep, items[i]), out); }
if (format) /* if requested, print information */
isr_rinfo(isrep, frq, body, head, minval);
fputc('\n', out); n++; /* terminate the output line and */
} /* count the reported ass. rule */
} /* if (target <= TT_MAXIMAL) .. else .. */
if (fflush(out) != 0) error(E_FWRITE, fn_out);
if (out != stdout) fclose(out);
out = NULL; /* close the output file */
MSG(stderr, "[%d %s(s)] done ", n,
(target == TT_RULE) ? "rule" : "set");
MSG(stderr, "[%.2fs].\n", SEC_SINCE(t));
#ifdef BENCH /* if benchmark version, */
ist_stats(istree); /* show the search statistics */
#endif /* (especially memory usage) */
/* --- clean up --- */
#ifndef NDEBUG /* if this is a debug version */
isr_delete(isrep, 0); /* the item set reporter, */
ist_delete(istree); /* the item set tree, */
if (tatree) tt_delete(tatree, 0); /* the transaction tree, */
if (tabag) tb_delete(tabag, 0); /* the transaction bag, */
ib_delete(ibase); /* and the item base */
#endif
#ifdef STORAGE /* if storage debugging */
showmem("at end of program"); /* check memory usage */
#endif
return 0; /* return 'ok' */
} /* main() */