attributeinfo.c

C语言的朴素贝叶斯分类器代码

#include "attributeInfo.h"

AttributeInfo *AttributeInfoTable::searchInfo(Description v)
{
	for(AttributeInfo *tmp=table[hash(v)]; tmp; tmp=tmp->next)
		if(v==tmp->value) return tmp;
	return NULL;
}

AttributeInfo *AttributeInfoTable::searchInfo(int hashVal, Description v)
{
	for(AttributeInfo *tmp=table[hashVal]; tmp; tmp=tmp->next)
		if(v==tmp->value) return tmp;
	return NULL;
}

AttributeInfoTable::AttributeInfoTable(
	SupervisedDataTable &tbl, int col, ClassInfoTable &cTbl)
{
	n_Class = cTbl.numOfClasses();
//	cTable  = &cTbl;
	sTable  = &tbl;
//	cSTable = new ClassSummary(cTbl);

	int row = tbl.numOfInstances();
	int classCol = tbl.numOfAttributes();
#ifdef DBG
	cout << "classCol="<<classCol<<endl;
#endif

	initHashTable();
	n_Values = 0;
	for(int r=0; r<row; r++)
	{
		Description currDescription = tbl.value(r, col);
		int hashValue = hash(currDescription);
		AttributeInfo *tmp = searchInfo(hashValue, currDescription);

		if(tmp==NULL) { // not found, new distinct value
			tmp = new AttributeInfo(currDescription, n_Class);
			tmp->next = table[hashValue];
			table[hashValue] = tmp;
			n_Values++; // count the number of distinct values
		}

		tmp->inc(cTbl.classLabel(tbl.value(r, classCol)));
	}
}

void AttributeInfo::print(ClassSummary &cSTbl)
{
#ifdef DBG
	cout << "==== AttributeInfo ==== >>"<< endl;
#endif
	cout << "P(" << value << "|Ci)";
	for(int c=0; c<n_Class; c++)
	{
#ifdef DBG
		cout << "P(" << value << "|" <<cSTbl.getDescription(c)
						<< ")=";
#endif
		cout << "\t" << count[c] << "/" << cSTbl.getCount(c);
	}
	cout << endl;
}

void AttributeInfoTable::print(ClassSummary *cSTable)
{
#ifdef DBG
	cout << "==== AttributeInfo Table ====" << endl;
#endif
	for(int r=0; r<MAX_HASH_SIZE; r++)
	{
		if(table[r]) {
#ifdef DBG
			cout << "[" << r <<"] " ;
#endif
			for(AttributeInfo *tmp=table[r]; tmp; tmp=tmp->next)
				tmp->print(*cSTable); // class summary table
		}
	}
}

void AttributeInfoTable::initHashTable()
{
	for(int r=0; r<MAX_HASH_SIZE; r++)
		table[r] = NULL;
}