rawdatatable.c

C语言的朴素贝叶斯分类器代码

#include "rawDataTable.h"

RawDataTable::RawDataTable(SupervisedDataTable &sDTbl):
	DataTable()
{
	copyExtended((DataTable)sDTbl);	
}

RawDataTable::RawDataTable(Source &src, SupervisedDataTable
								&sDTbl):DataTable(src, 1)
{
#ifdef DBG
	cout << "sDTbl.numOfAttr="<<sDTbl.numOfAttributes()  << ", " <<
					"src.numOfDesc="<<src.numOfDescriptions() << endl;
#endif

	ERROR1((sDTbl.numOfAttributes() ==
									(src.numOfDescriptions()-1)),
		"RawDataTable::constructor: the training data has a \
\n  number of columns different from the testing's");

#ifdef DBG
	cout << "RawDataTable::constructor shows the data\n";
	DataTable::print();
#endif
}

void RawDataTable::setClassField(int r, Description d)
{
	setValue( r, Col()-1, d);
}

void RawDataTable::print()
{
	for(int r=0; r<Row(); r++) {
		int c;
		for(c=0; c<Col()-2; c++) 
			cout << value(r, c) << " ";
		cout << " | " << value(r, c++) << " | ";
		if(value(r, c) == ERRORVALUE)
			cout << "UNKNOWN";
		else
			if(value(r,c) != value(r,c-1))
				cout << "ERROR:";
		cout << value(r, c) << " ";
		cout << endl;
	}
}

void RawDataTable::reportErrorRate()
{
	int	n_Errors = 0;
	int c = Col() - 1;

#ifdef DBG
	cout << "Total number of instances: " << Row() <<endl;
#endif

	for(int r=0; r<Row(); r++)
		if(value(r, c) == ERRORVALUE ||
			value(r,c) != value(r,c-1))
				n_Errors ++;

	cout << "The error rate: "
					<<n_Errors<<"/"<<Row()<<"="<<(float)n_Errors/Row()*100
					<<"%"<<endl;
}