sample.c

C5.0 决策树算法C实现,代码仅供参考使用

/*************************************************************************/
/*									 */
/*	Source code for use with See5/C5.0 Release 1.19			 */
/*	-----------------------------------------------			 */
/*		      Copyright RuleQuest Research 2003			 */
/*									 */
/*	This code is provided "as is" without warranty of any kind,	 */
/*	either express or implied.  All use is at your own risk.	 */
/*									 */
/*************************************************************************/


/*************************************************************************/
/*									 */
/*	Sample program to demonstrate the use of See5/C5.0 classifiers	 */
/*	--------------------------------------------------------------	 */
/*									 */
/*	Compilation:							 */
/*									 */
/*	    Unix: use an ANSI C compiler such as gcc and include	 */
/*		  the math library, e.g. gcc sample.c -lm		 */
/*									 */
/*	    Windows: compile as a console application with symbol	 */
/*		  "WIN32" defined					 */
/*									 */
/*	This program accepts three command-line options:		 */
/*									 */
/*	    -f <filestem>   specify the application name (required)	 */
/*	    -r		    use rulesets instead of decision trees	 */
/*	    -x		    use a similar format as saved by the	 */
/*			    cross-reference window			 */
/*									 */
/*	The program expects to find the following files:		 */
/*									 */
/*	    <filestem>.names (the application names file)		 */
/*									 */
/*	    <filestem>.rules or <filestem>.tree (the classifier	files	 */
/*		 generated by C5.0 or See5)			 	 */
/*									 */
/*	    <filestem>.costs (optional -- the  costs file)		 */
/*									 */
/*	    <filestem>.cases (with a format similar to a .data file, but */
/*		allowing classes to be given as '?' meaning 'unknown')	 */
/*									 */
/*	Please note: the names file must be exactly as it was when	 */
/*	the classifier was generated.					 */
/*									 */
/*	For each case in <filestem.cases>, the program prints the	 */
/*	given class and then the class predicted by the classifier	 */
/*	together with the confidence of the prediction.			 */
/*									 */
/*	Revised August 2003						 */
/*									 */
/*************************************************************************/

#include "defns.h"
#include "global.c"
#include "hooks.c"



/*************************************************************************/
/*									 */
/*	Main                                                             */
/*									 */
/*************************************************************************/


int main(int Argc, char *Argv[])
/*  ----  */
{
    FILE		*F;
    Description		Case;
    int			CaseNo=0, MaxClassLen=5, o;
    ClassNo		Predict, c;
    Boolean		XRefForm=false;
    extern String	OptArg, Option;

    /*  Process options  */

    while ( (o = ProcessOption(Argc, Argv, "f+xr")) )
    {
	switch (o)
	{
	case 'f':   FileStem = OptArg;
		    break;
	case 'r':   RULES = true;
		    break;
	case 'x':   XRefForm = true;
		    break;
	case '?':   printf("    **Unrecognised option %s\n", Option);
		    exit(1);
	}
    }

    /*  Read information on attribute names, values, and classes  */

    if ( ! (F = GetFile(".names", "r")) ) Error(NOFILE, Fn, "");

    GetNames(F);

    /*  Read the appropriate classifier file.  Call CheckFile() to
	determine the number of trials, then allocate space for
	trees or rulesets  */

    if ( RULES )
    {
	CheckFile(".rules", false);
	RuleSet = AllocZero(TRIALS+1, CRuleSet);

	ForEach(Trial, 0, TRIALS-1)
	{
	    RuleSet[Trial] = GetRules(".rules");
	}
    }
    else
    {
	CheckFile(".tree", false);
	Pruned = AllocZero(TRIALS+1, Tree);

	ForEach(Trial, 0, TRIALS-1)
	{
	    Pruned[Trial] = GetTree(".tree");
	}
    }

    /*  Close the classifier file and reset the file variable  */
    
    fclose(TRf);
    TRf = 0;
    
    /*  Now classify the cases in file <filestem>.cases.
	This has the same format as a .data file except that
	the class can be "?" to indicate that it is unknown.  */

    if ( XRefForm )
    {
	ForEach(c, 1, MaxClass)
	{
	    if ( (o = strlen(ClassName[c])) > MaxClassLen ) MaxClassLen = o;
	}

	printf("%-15s %*s   [Predicted]\n\n", "Case", -MaxClassLen, "Class");
    }
    else
    {
	printf("Case\t\tGiven\t\tPredicted\n %s\t\tClass\t\tClass\n\n",
		( LabelAtt ? "ID" : "No" ));
    }

    if ( ! (F = GetFile(".cases", "r")) ) Error(NOFILE, Fn, "");

    ClassSum = AllocZero(MaxClass+1, float);  /* used in classification */
    Vote     = AllocZero(MaxClass+1, float);  /* used with boosting */

    LineNo = 0;

    while ( (Case = GetDescription(F, false)) )
    {
	/*  For this case, find the class predicted by See5/C5.0 model  */

	Predict = Classify(Case);

	/*  Print either case label or number  */

	if ( LabelAtt )
	{
	    printf("%-15.15s ", SVal(Case,LabelAtt));
	}
	else
	{
	    printf("%4d\t\t", ++CaseNo);
	}

	/*  Print the result for this case in alternative formats  */

	if ( XRefForm )
	{
	    printf("%*s", -MaxClassLen, ClassName[Class(Case)]);
	    if ( Class(Case) != Predict )
	    {
		printf("   [%s]", ClassName[Predict]);
	    }
	    printf("\n");
	}
	else
	{
	    printf("%-15.15s %-15.15s [%.2f]\n",
		    ClassName[Class(Case)], ClassName[Predict], Confidence);
	}

	/*  Free the memory used by this case  */

	FreeLastCase(Case);
    }

    /*  Close the case file and free allocated memory  */
    
    fclose(F);
    FreeGlobals();
    
    return 0;
}