tract.h

apriori算法是数据挖掘的经典算法之1,其基于关联规则的思想.这是我的第2个收藏算法

/*----------------------------------------------------------------------
  File    : tract.h
  Contents: item and transaction management
  Author  : Christian Borgelt
  History : 18.11.2001 file created from file apriori.c
            28.12.2001 first version completed
            02.01.2001 ta_sort mapped to v_intsort
----------------------------------------------------------------------*/
#ifndef __TRACT__
#define __TRACT__
#ifndef NIMAPFN
#define NIMAPFN
#endif
#include "vecops.h"
#include "symtab.h"
#include "tfscan.h"

/*----------------------------------------------------------------------
  Preprocessor Definitions
----------------------------------------------------------------------*/
/* --- item appearance flags --- */
#define APP_NONE    0x00        /* item should be ignored */
#define APP_BODY    0x01        /* item may appear in rule body */
#define APP_HEAD    0x02        /* item may appear in rule head */
#define APP_BOTH    (APP_HEAD|APP_BODY)

/* --- error codes --- */
#define E_NONE        0         /* no error */
#define E_NOMEM     (-1)        /* not enough memory */
#define E_FOPEN     (-2)        /* cannot open file */
#define E_FREAD     (-3)        /* read error on file */
#define E_FWRITE    (-4)        /* write error on file */

#define E_ITEMEXP  (-16)        /* item expected */
#define E_DUPITEM  (-17)        /* duplicate item */
#define E_APPEXP   (-18)        /* appearance indicator expected */
#define E_UNKAPP   (-19)        /* unknown appearance indicator */
#define E_FLDCNT   (-20)        /* too many fields */

/*----------------------------------------------------------------------
  Type Definitions
----------------------------------------------------------------------*/
typedef struct {                /* --- an item --- */
  int     id;                   /* item identifier */
  int     frq;                  /* frequency in transactions */
  int     app;                  /* appearance indicator */
} ITEM;                         /* (item) */

typedef struct {                /* --- a transaction --- */
  int     cnt;                  /* number of items */
  int     items[1];             /* item identifier vector */
} TRACT;                        /* (transaction) */

typedef struct {                /* --- an itemset --- */
  TFSCAN  *tfscan;              /* table file scanner */
  char    chars[4];             /* special characters */
  NIMAP   *nimap;               /* name/identifier map */
  int     app;                  /* default appearance indicator */
  int     vsz;                  /* size of transaction buffer */
  int     cnt;                  /* number of items in transaction */
  int     *items;               /* items in transaction */
} ITEMSET;                      /* (item set) */

typedef struct {                /* --- a transaction set --- */
  ITEMSET *itemset;             /* underlying item set */
  int     max;                  /* maximum number of items per t.a. */
  int     vsz;                  /* size of transaction vector */
  int     cnt;                  /* number of transactions */
  TRACT   **tracts;             /* transaction vector */
} TASET;                        /* (transaction set) */

/*----------------------------------------------------------------------
  Item Set Functions
----------------------------------------------------------------------*/
extern ITEMSET*    is_create  (void);
extern void        is_delete  (ITEMSET *iset);
extern TFSCAN*     is_tfscan  (ITEMSET *iset);
extern void        is_chars   (ITEMSET *iset, const char *blanks,
                                              const char *fldseps,
                                              const char *recseps,
                                              const char *cominds);

extern int         is_cnt     (ITEMSET *iset);
extern int         is_item    (ITEMSET *iset, const char *name);
extern const char* is_name    (ITEMSET *iset, int item);
extern int         is_getfrq  (ITEMSET *iset, int item);
extern int         is_setfrq  (ITEMSET *iset, int item, int frq);
extern int         is_addfrq  (ITEMSET *iset, int item, int frq);
extern int         is_getapp  (ITEMSET *iset, int item);
extern int         is_setapp  (ITEMSET *iset, int item, int app);

extern int         is_readapp (ITEMSET *iset, FILE *file);
extern int         is_read    (ITEMSET *iset, FILE *file);

extern int         is_recode  (ITEMSET *iset, int minfrq, int dir,
                               int *map);
extern int         is_tsize   (ITEMSET *iset);
extern int*        is_tract   (ITEMSET *iset);

/*----------------------------------------------------------------------
  Transaction Functions
----------------------------------------------------------------------*/
extern void        ta_sort    (int *items, int n);
extern int         ta_unique  (int *items, int n);

/*----------------------------------------------------------------------
  Transaction Set Functions
----------------------------------------------------------------------*/
extern TASET*      tas_create  (ITEMSET *itemset);
extern void        tas_delete  (TASET *taset, int delis);
extern ITEMSET*    tas_itemset (TASET *taset);

extern int         tas_cnt     (TASET *taset);
extern int         tas_add     (TASET *taset, const int *items, int n);
extern int*        tas_tract   (TASET *taset, int index);
extern int         tas_tsize   (TASET *taset, int index);

extern void        tas_recode  (TASET *taset, int *map, int max);
extern void        tas_sort    (TASET *taset);
extern int         tas_occur   (TASET *taset, const int *items, int n);

#ifndef NDEBUG
extern void        tas_show    (TASET *taset);
#endif

/*----------------------------------------------------------------------
  Preprocessor Definitions
----------------------------------------------------------------------*/
#define is_tfscan(s)      ((s)->tfscan)

#define is_cnt(s)         nim_cnt((s)->nimap)
#define is_name(s,i)      nim_name(nim_byid((s)->nimap, i))
#define is_getfrq(s,i)    (((ITEM*)nim_byid((s)->nimap, i))->frq)
#define is_setfrq(s,i,f)  (((ITEM*)nim_byid((s)->nimap, i))->frq  = (f))
#define is_addfrq(s,i,f)  (((ITEM*)nim_byid((s)->nimap, i))->frq += (f))
#define is_getapp(s,i)    (((ITEM*)nim_byid((s)->nimap, i))->app)
#define is_setapp(s,i,a)  (((ITEM*)nim_byid((s)->nimap, i))->app  = (a))

#define is_tsize(s)       ((s)->cnt)
#define is_tract(s)       ((s)->items)

/*--------------------------------------------------------------------*/
#define ta_sort(v,n)      v_intsort(v,n)

/*--------------------------------------------------------------------*/
#define tas_itemset(s)    ((s)->itemset)
#define tas_cnt(s)        ((s)->cnt)
#define tas_max(s)        ((s)->max)

#define tas_tract(s,i)    ((s)->tracts[i]->items)
#define tas_tsize(s,i)    ((s)->tracts[i]->cnt)
#endif