stdafx.cpp

贝叶斯分类器设计

		for (i=0; i<ClassNum; i++){
			temp = ClassPreProbArray.ElementAt(i); 
			SqlInsertProb = SqlInsertProb + temp + ",";	
		}
		tempLength = SqlInsertProb.GetLength();
		tempLength = tempLength -1;
		SqlInsertProb = SqlInsertProb.Left(tempLength);
		SqlInsertProb = SqlInsertProb + ")";
		//AfxMessageBox(SqlInsertProb);
		CSim_tcApp::AdoDBObject.m_pConnClsDB->Execute((_bstr_t)SqlInsertProb,&RecordsAffected,adCmdText);
	
		//计算在每个类别条件下,每一个单词出现的后验概率
		//1.计算 分母(单词总数+整个矩阵的词频和)
		//单词总数为m_TrainFileDim
		float valFenMu=0.0;
		CString SqlGetWordFreq = "select sum(wd";
		_variant_t TotalFreq;
		
		for (i=1; i<=m_TrainFileDim; i++){
			_itoa( i, buffer, 10 );
			SqlGetWordFreq = SqlGetWordFreq + buffer +") from " + CurTrainSet;
			//AfxMessageBox(SqlGetWordFreq);
			CSim_tcApp::AdoDBObject.m_pRecordset = CSim_tcApp::AdoDBObject.m_pConnection->Execute((_bstr_t)SqlGetWordFreq,&RecordsAffected,adCmdText);
			TotalFreq = CSim_tcApp::AdoDBObject.m_pRecordset->GetCollect(vIndex);
			valFenMu = valFenMu + TotalFreq.bVal;
			SqlGetWordFreq = "select sum(wd";
		}
		valFenMu = valFenMu + m_TrainFileDim;
		//2.计算分子，分别计算每个单词在每个类别下的后验概率
		SqlGetWordFreq = "select sum(wd";
		int j=0;
		float valFenZi=0.0;
		float AWordProb;
		_variant_t vAWordFreq;
		CString SqlInsertWordFreq = "";
		for (i=1; i<=m_TrainFileDim; i++){
			for (j=0; j<ClassNum; j++){//单词i对应的各个类别的概率
				temp = ClassLabelArray.GetAt(j);
				_itoa(i, buffer, 10 );
				SqlGetWordFreq = SqlGetWordFreq + buffer +") from " + CurTrainSet + " where classlabel='" + temp + "'" ;
				//AfxMessageBox(SqlGetWordFreq);
				CSim_tcApp::AdoDBObject.m_pRecordset = CSim_tcApp::AdoDBObject.m_pConnection->Execute((_bstr_t)SqlGetWordFreq,&RecordsAffected,adCmdText);
				vAWordFreq = CSim_tcApp::AdoDBObject.m_pRecordset->GetCollect(vIndex);						
				valFenZi = valFenZi + vAWordFreq.bVal;
				valFenZi = valFenZi + 1;
				AWordProb = valFenZi / valFenMu;
				_gcvt(AWordProb, 10, buffer);
				ClassPreProbArray.SetAt(j,buffer);
				CSim_tcApp::AdoDBObject.m_pRecordset->Close();
				SqlGetWordFreq = "select sum(wd";
				valFenZi = 0.0;
			}
			SqlInsertWordFreq = "insert into " + ProbTableName + " values("; 
			for (j=0; j<ClassNum; j++){
				SqlInsertWordFreq = SqlInsertWordFreq + ClassPreProbArray.GetAt(j) + ",";
			}
			SqlInsertWordFreq = SqlInsertWordFreq.Left(SqlInsertWordFreq.GetLength()-1);
			SqlInsertWordFreq = SqlInsertWordFreq + ")";
			//AfxMessageBox(SqlInsertWordFreq);
			CSim_tcApp::AdoDBObject.m_pConnClsDB->Execute((_bstr_t)SqlInsertWordFreq,&RecordsAffected,adCmdText);
			SqlGetWordFreq = "select sum(wd";
			SqlInsertWordFreq = "";

		}
		AfxMessageBox("成功了也！");
	}
	else {
		return;
	}
}

//Bayes的贝努利模型训练方法
void BernulliBayesTrain(CString CurTrainSet)
{
	_variant_t RecordsAffected;
	_variant_t vAttrNum;
	_variant_t vClassNum;
	_variant_t vClassLabel;
	_variant_t vIndex = (long)0;

	CString SqlCheckTrainSetDim="";
	int m_TrainFileDim;
	
	//在classification数据库中建立表存放中间求出的概率结果
	//1.在classification数据库中创建新表,用于存放中间结果
	CString SqlCreateProbTable = "";
	CString SqlCheckTrained = "";
	CString ProbTableName = "",temp="";
	ProbTableName = CurTrainSet + "_ProbResult";
	
	_variant_t vCheckResult;
	int CheckResult;
	//检查是否已经训练过
	SqlCheckTrained = "select count(name) from sysobjects where xtype='u' and name<>'dtproperties' and name='" +\
		ProbTableName + "'";
	//AfxMessageBox(SqlCheckTrained);
	CSim_tcApp::AdoDBObject.m_pRecordset = CSim_tcApp::AdoDBObject.m_pConnClsDB->Execute((_bstr_t)SqlCheckTrained,&RecordsAffected,adCmdText);
	vCheckResult = CSim_tcApp::AdoDBObject.m_pRecordset->GetCollect(vIndex);
	CheckResult = vCheckResult.lVal;
	CSim_tcApp::AdoDBObject.m_pRecordset->Close();

	if (CheckResult == 0){
	
	//2.维数(单词的个数)
		SqlCheckTrainSetDim = "select count(name) from syscolumns where syscolumns.id in (select id from sysobjects where name = '" + CurTrainSet + "')"; //查询属性个数
		CSim_tcApp::AdoDBObject.m_pRecordset = CSim_tcApp::AdoDBObject.m_pConnection->Execute((_bstr_t)SqlCheckTrainSetDim,&RecordsAffected,adCmdText);
		vAttrNum = CSim_tcApp::AdoDBObject.m_pRecordset->GetCollect(vIndex);
		m_TrainFileDim = vAttrNum.lVal;
		m_TrainFileDim = m_TrainFileDim - 2;
		CSim_tcApp::AdoDBObject.m_pRecordset->Close();
		
		//2.类别的名称
		StringArray ClassLabelArray;
		CString SqlGetClassLabel = "";
		CString ClassLabel="";
		
		// 查询类别个数
		SqlGetClassLabel = "select count(distinct classlabel) from   " + CurTrainSet; 
		CSim_tcApp::AdoDBObject.m_pRecordset = CSim_tcApp::AdoDBObject.m_pConnection->Execute((_bstr_t)SqlGetClassLabel,&RecordsAffected,adCmdText);
		vClassNum = CSim_tcApp::AdoDBObject.m_pRecordset->GetCollect(vIndex);
		int ClassNum = vClassNum.lVal;
		CSim_tcApp::AdoDBObject.m_pRecordset->Close();
		ClassLabelArray.SetSize(ClassNum,5);
		//查询类别名称
		SqlGetClassLabel = "select distinct classlabel from   " + CurTrainSet; 
		CSim_tcApp::AdoDBObject.m_pRecordset = CSim_tcApp::AdoDBObject.m_pConnection->Execute((_bstr_t)SqlGetClassLabel,&RecordsAffected,adCmdText);
		int i=0;
		while(!CSim_tcApp::AdoDBObject.m_pRecordset->adoEOF){
			vClassLabel = (_bstr_t)CSim_tcApp::AdoDBObject.m_pRecordset->GetCollect(vIndex);
			ClassLabel = vClassLabel.bstrVal;
			ClassLabel.TrimLeft();
			ClassLabel.TrimRight();
			ClassLabelArray.SetAt(i,ClassLabel);
			i++;
			CSim_tcApp::AdoDBObject.m_pRecordset->MoveNext();
		}
		CSim_tcApp::AdoDBObject.m_pRecordset->Close();
		ClassLabelArray.FreeExtra();

	
		SqlCreateProbTable = "Create table " + ProbTableName + "(rowid int identity(0,1) not null primary key, ";
		for (i=0; i<ClassNum; i++){
			temp = ClassLabelArray.GetAt(i); 
			SqlCreateProbTable = SqlCreateProbTable + temp + " float,";
		}
		int tempLength = SqlCreateProbTable.GetLength();
		tempLength = tempLength -1;
		SqlCreateProbTable = SqlCreateProbTable.Left(tempLength);
		SqlCreateProbTable = SqlCreateProbTable + ")";
		//AfxMessageBox(SqlCreateProbTable);
		CSim_tcApp::AdoDBObject.m_pConnClsDB->Execute((_bstr_t)SqlCreateProbTable,&RecordsAffected,adCmdText);
	
		//计算每一个类别出现的先验概率
		_variant_t vAllDocNum;
		_variant_t vAClassDocNum;
		float AllDocNum,AClassDocNum=0.0;
		CString StrClassDocNum,strAClassDocNum;
		CString SqlGetAllDocNum = "", SqlGetAClassDocNum="";
		SqlGetAllDocNum = "select count(*) from " + CurTrainSet;
		//AfxMessageBox(SqlGetAllDocNum);
		CSim_tcApp::AdoDBObject.m_pRecordset = CSim_tcApp::AdoDBObject.m_pConnection->Execute((_bstr_t)SqlGetAllDocNum,&RecordsAffected,adCmdText);
		vAllDocNum = (_bstr_t)CSim_tcApp::AdoDBObject.m_pRecordset->GetCollect(vIndex);
		StrClassDocNum = vAllDocNum.bstrVal;
		AllDocNum = atof(StrClassDocNum);
		CSim_tcApp::AdoDBObject.m_pRecordset->Close();
	
		StringArray ClassPreProbArray;
		ClassPreProbArray.SetSize(ClassNum,5);

		float tempProb=0.0;
		char buffer[50];
		for (i=0; i<ClassNum; i++){
			temp = ClassLabelArray.ElementAt(i);
			SqlGetAClassDocNum = "select count(*) from " + CurTrainSet + " where classlabel='" + temp + "'";
			//AfxMessageBox(SqlGetAClassDocNum);
			CSim_tcApp::AdoDBObject.m_pRecordset = CSim_tcApp::AdoDBObject.m_pConnection->Execute((_bstr_t)SqlGetAClassDocNum,&RecordsAffected,adCmdText);
			vAClassDocNum = (_bstr_t)CSim_tcApp::AdoDBObject.m_pRecordset->GetCollect(vIndex);
			strAClassDocNum = vAClassDocNum.bstrVal;
			AClassDocNum = atof(strAClassDocNum);
			tempProb = AClassDocNum / AllDocNum;
			_gcvt(tempProb, 10, buffer);
			ClassPreProbArray.SetAt(i,buffer);
			CSim_tcApp::AdoDBObject.m_pRecordset->Close();
		}
		CString SqlInsertProb = "insert into " + ProbTableName + " values (";
	
		for (i=0; i<ClassNum; i++){
			temp = ClassPreProbArray.ElementAt(i); 
			SqlInsertProb = SqlInsertProb + temp + ",";	
		}
		tempLength = SqlInsertProb.GetLength();
		tempLength = tempLength -1;
		SqlInsertProb = SqlInsertProb.Left(tempLength);
		SqlInsertProb = SqlInsertProb + ")";
		//AfxMessageBox(SqlInsertProb);
		CSim_tcApp::AdoDBObject.m_pConnClsDB->Execute((_bstr_t)SqlInsertProb,&RecordsAffected,adCmdText);
	
		//计算在每个类别条件下,每一个单词出现的后验概率
		//1.计算 分母(单词总数+整个矩阵的词频和)
		//单词总数为m_TrainFileDim
		float valFenMu=0.0;
		CString SqlGetWordFreq = "select sum(wd";
		_variant_t TotalFreq;
		
		for (i=1; i<=m_TrainFileDim; i++){
			_itoa( i, buffer, 10 );
			SqlGetWordFreq = SqlGetWordFreq + buffer +") from " + CurTrainSet;
			//AfxMessageBox(SqlGetWordFreq);
			CSim_tcApp::AdoDBObject.m_pRecordset = CSim_tcApp::AdoDBObject.m_pConnection->Execute((_bstr_t)SqlGetWordFreq,&RecordsAffected,adCmdText);
			TotalFreq = CSim_tcApp::AdoDBObject.m_pRecordset->GetCollect(vIndex);
			valFenMu = valFenMu + TotalFreq.bVal;
			SqlGetWordFreq = "select sum(wd";
		}
		valFenMu = valFenMu + m_TrainFileDim;
		//2.计算分子，分别计算每个单词在每个类别下的后验概率
		SqlGetWordFreq = "select sum(wd";
		int j=0;
		float valFenZi=0.0;
		float AWordProb;
		_variant_t vAWordFreq;
		CString SqlInsertWordFreq = "";
		for (i=1; i<=m_TrainFileDim; i++){
			for (j=0; j<ClassNum; j++){//单词i对应的各个类别的概率
				temp = ClassLabelArray.GetAt(j);
				_itoa(i, buffer, 10 );
				SqlGetWordFreq = SqlGetWordFreq + buffer +") from " + CurTrainSet + " where classlabel='" + temp + "'" ;
				//AfxMessageBox(SqlGetWordFreq);
				CSim_tcApp::AdoDBObject.m_pRecordset = CSim_tcApp::AdoDBObject.m_pConnection->Execute((_bstr_t)SqlGetWordFreq,&RecordsAffected,adCmdText);
				vAWordFreq = CSim_tcApp::AdoDBObject.m_pRecordset->GetCollect(vIndex);						
				valFenZi = valFenZi + vAWordFreq.bVal;
				valFenZi = valFenZi + 1;
				AWordProb = valFenZi / valFenMu;
				_gcvt(AWordProb, 10, buffer);
				ClassPreProbArray.SetAt(j,buffer);
				CSim_tcApp::AdoDBObject.m_pRecordset->Close();
				SqlGetWordFreq = "select sum(wd";
				valFenZi = 0.0;
			}
			SqlInsertWordFreq = "insert into " + ProbTableName + " values("; 
			for (j=0; j<ClassNum; j++){
				SqlInsertWordFreq = SqlInsertWordFreq + ClassPreProbArray.GetAt(j) + ",";
			}
			SqlInsertWordFreq = SqlInsertWordFreq.Left(SqlInsertWordFreq.GetLength()-1);
			SqlInsertWordFreq = SqlInsertWordFreq + ")";
			//AfxMessageBox(SqlInsertWordFreq);
			CSim_tcApp::AdoDBObject.m_pConnClsDB->Execute((_bstr_t)SqlInsertWordFreq,&RecordsAffected,adCmdText);
			SqlGetWordFreq = "select sum(wd";
			SqlInsertWordFreq = "";

		}
		AfxMessageBox("成功了也！");
	}
	else {
		return;
	}
}