gramod.h

数据挖掘中的一算法 ines算法 c下实现的。适合初学习数据挖掘者借鉴

/*----------------------------------------------------------------------
  File    : gramod.h
  Contents: graphical model management (based on directed graphs)
  Author  : Christian Borgelt
  History : 23.02.1996 file created as ascons.h from attset.h
            09.04.1996 function gm_rand modified (parameter randfn())
            15.11.1996 description modes added, gm_dmode extended
            27.11.1996 adapted to floating point counters
            28.02.1997 adapted to prob./poss. tree types
            01.03.1997 function gm_exec added
            08.02.1998 function gm_parse transferred from parse.h
            01.03.1998 adapted to extended prob./poss. tree structure
            05.04.1999 parameters 'attid' and 'conid' added to gm_check
            30.04.1999 definition of GM_ALL added
            08.02.2002 adapted to changed module ptree
            11.03.2002 parse flag GM_DFLT (create default network) added
            30.03.2002 function gm_evalx added (from sian.c)
            04.04.2002 functions gm_consum and gm_parsum added
            11.04.2002 function gm_tree added
            16.07.2002 parameter 'delnodes' added to function gm_clear
----------------------------------------------------------------------*/
#ifndef __GRAMOD__
#define __GRAMOD__
#if defined(GM_PARSE) && !defined(PT_PARSE)
#define PT_PARSE
#endif
#include "parse.h"
#if defined(GM_EVAL) || defined(GM_INDUCE)
#include "table.h"
#else
#include "attset.h"
#endif
#include "ptree.h"

/*----------------------------------------------------------------------
  Preprocessor Definitions
----------------------------------------------------------------------*/
/* --- types --- */
#define GM_PROB     PT_PROB     /* probabilistic (sum) */
#define GM_POSS     PT_POSS     /* possibilistic (maximum) */
#define GM_AUTO     PT_AUTO     /* determine type from input file */

/* --- description/parse modes --- */
#define GM_TITLE    PT_TITLE    /* print a title (as a comment) */
#define GM_INFO     PT_INFO     /* print add. info. (as a comment) */
#define GM_FULL     PT_FULL     /* print full trees (all branches) */
#define GM_REL      PT_REL      /* print relative frequencies */
#define GM_CDDS     0x0100      /* parse candidate lists */
#define GM_ALL      0x0200      /* parse a complete set of trees */
#define GM_DFLT     0x0400      /* create a default network */

/* --- search methods --- */
#define GM_NONE       0         /* no search, only parameter estim. */
#define GM_OWST       1         /* optimum weight spanning tree */
#define GM_EXTST      2         /* spanning tree extension */
#define GM_TOPORD     3         /* cond. sel. on topological order */
#define GM_NOLOOP     4         /* cond. sel. without directed loops */
#define GM_UNRES      5         /* cond. sel. without restrictions */
#define GM_SIAN       6         /* hypertree simulated annealing */
#define GM_UNKNOWN    7         /* unknown search method */

/* --- other values --- */
#define GM_ERROR    PT_ERROR    /* to indicate an evaluation error */

/*----------------------------------------------------------------------
  Type Definitions
----------------------------------------------------------------------*/
typedef struct {                /* --- attribute specific data --- */
  ATT    *att;                  /* corresponding attribute */
  int    cddvsz;                /* size of candidate attribute vector */
  int    cddcnt;                /* number of candidate attributes */
  int    *cdds;                 /* candidate attributes */
  int    depvsz;                /* size of dependent attribute vector */
  int    depcnt;                /* number of dependent attributes */
  int    *deps;                 /* dependent attributes */
  PTREE  *ptree;                /* current prob./poss. tree */
  int    pos;                   /* position in topological order */
  float  prob;                  /* probability used in gm_execx() */
  double eval;                  /* evaluation result */
} GMATT;                        /* (attribute specific data) */

typedef struct {                /* --- graphical model --- */
  ATTSET *attset;               /* underlying attribute set */
  int    type;                  /* type of model (prob./poss.) */
  int    attcnt;                /* number of attributes */
  int    concnt;                /* number of conditions */
  double tplcnt;                /* number of tuples for norm. */
  float  lcorr;                 /* Laplace correction */
  int    *order;                /* topological order of attributes */
  int    *buf;                  /* buffer for execution and parsing */
  double eval;                  /* evaluation result */
  GMATT  atts[1];               /* vector of attribute specific data */
} GRAMOD;                       /* (attribute set conditions) */

/*----------------------------------------------------------------------
  Functions
----------------------------------------------------------------------*/
extern GRAMOD* gm_create  (ATTSET *attset, int type);
extern void    gm_delete  (GRAMOD *gm, int delas);

extern ATTSET* gm_attset  (GRAMOD *gm);
extern int     gm_attcnt  (GRAMOD *gm);
extern int     gm_type    (GRAMOD *gm);
extern double  gm_setnorm (GRAMOD *gm, double tplcnt);
extern double  gm_getnorm (GRAMOD *gm);
extern float   gm_setlc   (GRAMOD *gm, float lcorr);
extern float   gm_getlc   (GRAMOD *gm);
extern int     gm_consum  (GRAMOD *gm);
extern int     gm_parsum  (GRAMOD *gm);

extern int     gm_cddcnt  (GRAMOD *gm, int attid);
extern int     gm_cddadd  (GRAMOD *gm, int attid, int cddid);
extern void    gm_cddrem  (GRAMOD *gm, int attid, int index);
extern int     gm_cddid   (GRAMOD *gm, int attid, int index);

extern int     gm_concnt  (GRAMOD *gm, int attid);
extern int     gm_conadd  (GRAMOD *gm, int attid, int conid);
extern void    gm_conrem  (GRAMOD *gm, int attid, int index);
extern int     gm_conid   (GRAMOD *gm, int attid, int index);
extern ATT*    gm_con     (GRAMOD *gm, int attid, int index);

extern int     gm_depcnt  (GRAMOD *gm, int attid);
extern int     gm_depid   (GRAMOD *gm, int attid, int index);
extern ATT*    gm_dep     (GRAMOD *gm, int attid, int index);

extern PTREE*  gm_ptree   (GRAMOD *gm, int attid);
extern void    gm_clear   (GRAMOD *gm, int delnodes);
extern int     gm_aggr    (GRAMOD *gm, int distuv);
extern int     gm_check   (GRAMOD *gm, int attid, int conid);
extern int     gm_rand    (GRAMOD *gm, double randfn(void));
extern double  gm_exec    (GRAMOD *gm);
extern int     gm_execx   (GRAMOD *gm, double res[]);

#ifdef GM_EVAL
extern int     gm_estim   (GRAMOD *gm, TABLE *table, int distuv);
extern double  gm_eval    (GRAMOD *gm, int measure, double *params);
extern double  gm_evalx   (GRAMOD *gm, TABLE *table, double res[]);
extern double  gm_local   (GRAMOD *gm, int lwise, double minimp);
extern CCHAR*  gm_sname   (int search);
#endif
#ifdef GM_INDUCE
extern double  gm_tree    (GRAMOD *gm, TABLE *table, int search,
                           int measure, double *params, int maxtest,
                           int verbose);
extern double  gm_csel    (GRAMOD *gm, TABLE *table, int search,
                           int measure, double *params, double minimp,
                           int lwise, double mrgimp, int maxcon,
                           int maxtest, int verbose);
extern double  gm_sian    (GRAMOD *gm, TABLE *table, int trials,
                           int maxsz, double keep, double frac,
                           double ppwgt, int distuv,
                           double randfn (void), int verbose);
#endif
extern int     gm_desc    (GRAMOD *gm, FILE *file, int mode,int maxlen);
#ifdef GM_PARSE
extern GRAMOD* gm_parse   (ATTSET *attset, SCAN *scan, int type);
#endif

/*----------------------------------------------------------------------
  Preprocessor Definitions
----------------------------------------------------------------------*/
#define gm_attset(c)      ((c)->attset)
#define gm_attcnt(c)      ((c)->attcnt)
#define gm_type(c)        ((c)->type)
#define gm_getnorm(c)     ((c)->tplcnt)
#define gm_getlc(c)       ((c)->lcorr)

#define gm_cddcnt(c,a)    ((c)->atts[a].cddcnt)
#define gm_cddid(c,a,i)   ((c)->atts[a].cdds[i])

#define gm_concnt(c,a)    ((a) < 0) ? (c)->concnt \
                                    : pt_concnt((c)->atts[a].ptree)
#define gm_conid(c,a,i)   pt_conid((c)->atts[a].ptree, i)
#define gm_con(c,a,i)     pt_con((c)->atts[a].ptree, i)
#define gm_ptree(c,a)     ((c)->atts[a].ptree)

#define gm_depcnt(c,a)    ((c)->atts[a].depcnt)
#define gm_depid(c,a,i)   ((c)->atts[a].deps[i])
#define gm_dep(c,a,i)     (as_att((c)->attset, (c)->atts[a].deps[i]))

#define gm_tree(g,t,s,m,p,x,v)  gm_csel(g,t,s,m,p,0,-1,0,2,x,v)
#endif