datasource.h

关联分类算法采用贪心算法发现高质量分类规则

// DataSource.h: interface for the DataSource class.
//
//////////////////////////////////////////////////////////////////////

#if !defined(AFX_DATASOURCE_H__352E590A_63B9_4D5C_BB3C_E1D7BC7B8C2B__INCLUDED_)
#define AFX_DATASOURCE_H__352E590A_63B9_4D5C_BB3C_E1D7BC7B8C2B__INCLUDED_

#if _MSC_VER > 1000
#pragma once
#endif // _MSC_VER > 1000

#include <list>

class DataSource  
{
public:
	DataSource();
	virtual ~DataSource();

	bool		ReadData(CString filename);
	bool		ReadRules(CString filename);
	void		ResetAllWeights();
	double		TotalWeights();
	double		TotalWeights(const std::list<TUPLE*>& list);

	double		EvaluateRule(const RULE& rule,bool expected_error=true) const;
	double		EvaluateRule(const RULE& rule,const std::list<TUPLE*>& P,const std::list<TUPLE*>& N,bool expected_error=true) const;
	void		EvaluateRule(const RULE& rule,int& sup,double& conf,double& lift);

	void		GenAllLiterals(std::list<LITERAL>& llist);
	double		EvaluateLiteral(const LITERAL& lit,const std::list<TUPLE*>& P,const std::list<TUPLE*>& N) const;
	double		EvaluateLiteral(double nP,double nN,double nP_,double nN_) const;
	void		GenRulesFOIL(int iClass,RULESET& Rules);
	int			GenRulesFOIL(CString outputfilename);

	int			Classify(const TUPLE& tuple,int& iRule) const;
	double		Classify(CString outputfilename) const;

	void		EvaluateRules(CString outputfile);
	void		IntepretRules(CString rule_semantics_file,CString outputfilename) const;

	void		GenAccrCurve(CString datafile,CString rulefile,CString outputfile);

protected:
	DESCRIPTION	m_Descript;			//the description
	
	int			m_nTuples;			//number of tuples
	TUPLE*		m_Tuples;

	int			m_nRules;
	RULE*		m_Rules;

	int*		m_nTupleInClass;

	std::list<RULE>	m_RuleStack;
};

#endif // !defined(AFX_DATASOURCE_H__352E590A_63B9_4D5C_BB3C_E1D7BC7B8C2B__INCLUDED_)