defns.h

这是数据挖掘算法的cubist算法的具体实现.运行在dos下.

	 {
	    Attribute	Tested;		/* split attribute for KD-tree */
	    ContValue	Cut,		/* threshold for continuous atts */
			MinDRef,	/* min reference distance */
			MaxDRef;	/* max ditto */
	    ItemNo	IFp, ILp;	/* first and last item at leaf */
	    Index	*SubIndex;	/* subtrees */
	 }
	 IndexRec;


typedef  struct _nnrec
	 {
	    int		BestI[MAXN];	/* numbers of best instances */
	    float	BestD[MAXN],	/* distances to best instances */
			*WorstBest,	/* points to worst BestD */
			*AttMinD;	/* min attribute distance from case */
	 }
	 NNEnvRec, *NNEnv;


typedef  union	 _def_val
	 {
	    String	_s_val;		/* att val for comparison */
	    ContValue	_n_val;		/* number for arith */
	 }
	 DefVal;

typedef  struct  _def_elt
	 {
	    short	_op_code;	/* type of element */
	    DefVal	_operand;	/* string or numeric value */
	 }
	 DefElt, *Definition;

typedef  struct  _elt_rec
	 {
	    int		Fi,		/* index of first char of element */
			Li;		/* last ditto */
	    char	Type;		/* 'B', 'S', or 'N' */
	 }
	 EltRec;

#define	 DefOp(DE)	DE._op_code
#define	 DefSVal(DE)	DE._operand._s_val
#define	 DefNVal(DE)	DE._operand._n_val

#define	 OP_ATT			 0	/* opcodes */
#define	 OP_NUM			 1
#define	 OP_STR			 2
#define	 OP_MISS		 3
#define	 OP_AND			10
#define	 OP_OR			11
#define	 OP_EQ			20
#define	 OP_NE			21
#define	 OP_GT			22
#define	 OP_GE			23
#define	 OP_LT			24
#define	 OP_LE			25
#define	 OP_SEQ			26
#define	 OP_SNE			27
#define	 OP_PLUS		30
#define	 OP_MINUS		31
#define	 OP_UMINUS		32
#define	 OP_MULT		33
#define	 OP_DIV			34
#define	 OP_MOD			35
#define	 OP_POW			36
#define	 OP_SIN			40
#define	 OP_COS			41
#define	 OP_TAN			42
#define	 OP_LOG			43
#define	 OP_EXP			44
#define	 OP_INT			45
#define	 OP_END			99


/*************************************************************************/
/*									 */
/*		Function prototypes					 */
/*									 */
/*************************************************************************/


Boolean	    ReadName(FILE *f, String s, int n, char ColonOpt);
void	    GetNames(FILE *Nf);
void	    ExplicitAtt(FILE *Nf);
int	    Which(String Val, String *List, int First, int Last);
void	    FreeNamesData();
int	    InChar(FILE *f);

void	    ImplicitAtt(FILE *Nf);
void	    ReadDefinition(FILE *f);
void	    Append(char c);
Boolean	    Expression();
Boolean	    Conjunct();
Boolean	    SExpression();
Boolean	    AExpression();
Boolean	    Term();
Boolean	    Factor();
Boolean	    Primary();
Boolean	    Atom();
Boolean	    Find(String S);
int	    FindOne(String *Alt);
Attribute   FindAttName();
void	    DefSyntaxError(String Msg);
void	    DefSemanticsError(int Fi, String Msg, int OpCode);
void	    Dump(char OpCode, ContValue F, String S, int Fi);
void	    DumpOp(char OpCode, int Fi);
Boolean	    UpdateTStack(char OpCode, ContValue F, String S, int Fi);
AttValue    EvaluateDef(Definition D, Description Case);

void	    GetData(FILE *Df, Boolean Train, Boolean AllowUnknownTarget);
Boolean	    ReplaceUnknowns(Description Case, Boolean *AttMsg);
Description GetDescription(FILE *Df, Boolean Train);
int	    StoreIVal(String S);
void	    CheckValue(Description Case, Attribute Att);
void	    FreeCases(Description *Case, ItemNo MaxCase);
void	    FreeCase(Description DVec);

void	    CheckFile(String Extension, Boolean Write);
void	    ReadFilePrefix(String Extension);
void	    ReadHeader();
RRuleSet    *GetCommittee(String Extension);
RRuleSet    InRules();
CRule	    InRule();
Condition   InCondition();
void	    FreeCttee(RRuleSet *Cttee);
int	    ReadProp(char *Delim);
String	    RemoveQuotes(String S);
Set	    MakeSubset(Attribute Att);
void	    BinRecoverDiscreteNames();
RRuleSet    BinInRules();
void	    StreamIn(String S, int n);

void	    ReleaseRule(CRule R);
void	    PrintRules(RRuleSet RS, String Msg);
void	    PrintRule(CRule R);
void	    PrintCondition(Condition C);
Boolean	    Satisfies(Description CaseDesc, Condition OneCond);

float	    PredictValue(RRuleSet *Cttee, Description CaseDesc);
float	    RuleSetPrediction(RRuleSet RS, Description CaseDesc);
Boolean	    Matches(CRule R, Description Case);

void	    InitialiseInstances(RRuleSet *Cttee);
float	    NNEstimate(RRuleSet *Cttee, Description Case, NNEnv E);
float	    Distance(Description Case1, Description Case2, float Thresh);
void	    CheckDistance(Description Case, ItemNo Saved, NNEnv E);
void	    FindNearestNeighbors(Description Case, NNEnv E);
float	    AverageNeighbors(RRuleSet *Cttee, Description Case, NNEnv E);
Index	    BuildIndex(Attribute Att, ItemNo Fp, ItemNo Lp);
void	    ScanIndex(Description Case, Index Node, float MinD, NNEnv E);
void	    SwapInstance(ItemNo A, ItemNo B);
void	    FreeIndex(Index Node);

char	    ProcessOption(int Argc, char *Argv[], char *Options);
void	    *Pmalloc(size_t Bytes);
void	    *Prealloc(void *Present, size_t Bytes);
void	    *Pcalloc(size_t Number, unsigned int Size);
void	    FreeVector(void **V, int First, int Last);

void	    Error(int ErrNo, String S1, String S2);
FILE	    *GetFile(String Extension, String RW);
int	    Denominator(ContValue Val);
int	    FracBase(Attribute Att);
int	    GetInt(String S, int N);
int	    DateToDay(String DS);
void	    DayToDate(int Day, String Date);
int	    TimeToSecs(String TS);
void	    SecsToTime(int Secs, String Time);
void	    SetTSBase(int y);
int	    TStampToMins(String TS);
void	    CValToStr(ContValue CV, Attribute Att, String DS);

#ifdef WIN32
double	    posrint(double);
double	    rint(double);
#endif


/*************************************************************************/
/*									 */
/*		Text strings						 */
/*									 */
/*************************************************************************/


#define	 T_IgnoreNATarget	"*** Ignoring instances with N/A target value\n"
#define	 T_IgnoreBadTarget	"*** Ignoring instances with unknown or N/A"\
					" target value\n"
#define	 T_NoCases		"*** No instances with known target values\n"

#define	 T_ReplaceUnknowns	"\n    Replacing unknown attribute values:\n"
#define	 T_NoAppVals		"ignoring (no applicable values)"
#define	 T_By			"by"
#define	 T_Rule			"Rule"
#define	 TX_RInfo(c,p,m,l,h,e)	": [%d cases, mean %.*f, range %.7g to %.7g, "\
				"est err %.*f]\n\n",c,p,m,l,h,p,e
#define	 T_If			"if"
#define	 T_Then			"then"
#define	 T_ElementOf		"in"
#define	 T_InRange		"in"
#define	 T_IsUnknown		" is unknown\n"
#define	 TX_Line(l,f)		"\n*** line %d of `%s': ", l, f
#define	 E_NOFILE(f,e)		"cannot open file %s%s\n", f, e
#define	 E_BADATTNAME		"`:' or `:=' expected after attribute name"\
					" `%s'\n"
#define	 E_EOFINATT		"unexpected eof while reading attribute `%s'\n"
#define	 E_SINGLEATTVAL(a,v)	"attribute `%s' has only one value `%s'\n",\
					a, v
#define	 E_DUPATTNAME		"multiple attributes with name `%s'\n"
#define	 E_CWTATTERR		"case weight attribute must be continuous\n"
#define	 E_BADATTVAL(v,a)	"bad value of `%s' for attribute `%s'\n", v, a
#define	 E_BADNUMBER(a)		"value of `%s' changed to `?'\n", a
#define	 E_NOMEM		"unable to allocate sufficient memory\n"
#define	 E_TOOMANYVALS(a,n)	"too many values for attribute `%s'"\
					" (max %d)\n", a, n
#define	 E_BADDISCRETE		"bad number of discrete values for attribute"\
					" `%s'\n"
#define	 E_NOTARGET		"target attribute `%s' not found\n"
#define	 E_BADTARGET		"target attribute `%s' is not numeric\n"
#define	 E_LONGNAME		"overlength name: check data file formats\n"
#define	 E_HITEOF		"unexpected end of file\n"
#define	 E_MISSNAME		"missing name or value before `%s'\n"
#define	 E_BADTSTMP(d,a)	"bad timestamp `%s' for attribute `%s'\n", d, a
#define	 E_BADDATE(d,a)		"bad date `%s' for attribute `%s'\n", d, a
#define	 E_BADTIME(d,a)		"bad time `%s' for attribute `%s'\n", d, a
#define	 E_UNKNOWNATT		"unknown attribute name `%s'\n"
#define	 E_BADDEF1(a,s,x)	"in definition of attribute `%s':\n"\
					"\tat `%.12s': expect %s\n", a, s, x
#define	 E_BADDEF2(a,s,x)	"in definition of attribute `%s':\n"\
					"\t`%s': %s\n", a, s, x
#define	 E_SAMEATT(a,b)		"attribute `%s' is identical to attribute"\
					" `%s'\n", a, b
#define	 E_BADDEF3		"cannot define target attribute `%s'\n"
#define	 E_BADDEF4		"target attribute appears in definition"\
					" of attribute `%s':\n"
#define	 EX_MODELFILE(f)	"file %s incompatible with .names file\n", f
#define	 E_MFATT		"undefined or excluded attribute"
#define	 E_MFATTVAL		"undefined attribute value"
#define	 E_MFEOF		"unexpected eof"
#define	 T_ErrorLimit		"Error limit exceeded\n"

#ifdef _CONSOLE
#define	 finite(x)	_finite(x)
#endif