char* DecisionTreeClassifier::findMajorityClass(const std::vector<attr_set*>* dec_set)
{
char* major_class = new char[256];
int freq = 0;
attr_set* attset;
for(std::vector<attr_set*>::const_iterator it1=dec_set->begin(); it1 != dec_set->end(); it1++)
{
attset = (*it1);
string str_class("Class");
attr_set::iterator attr_it = attset->find(str_class);
if(attr_it != attset->end())
{
string str_attr(attr_it->second);
freq = class_freq_map[str_attr];
class_freq_map[str_attr] = freq++;
}
}
for(std::tr1::unordered_map<string,int>::iterator it2=class_freq_map.begin();it2 != class_freq_map.end(); it2++)
{
int max = 0;
if(it2->second > max)
{
strcpy(major_class, (it2->first).c_str());
max = it2->second;
}
}
return major_class;
}
double DecisionTreeClassifier::impurityEval1(const std::vector<attr_set*>* dec_set)
{
double entropy = 0.0;
double max = 0.0;
double total = 0.0;
double prob = 0.0;
class_freq_map.clear();
for(std::vector<attr_set*>::const_iterator it1=dec_set->begin(); it1 != dec_set->end(); it1++)
{
string str_class("Class");
attr_set::iterator attr_it = (*it1)->find(str_class);
if(attr_it != (*it1)->end())
{
string str_attr(attr_it->second);
int freq =class_freq_map[str_attr];
class_freq_map[str_attr] = ++freq;
}
}
for(std::tr1::unordered_map<string, int>::iterator it2=class_freq_map.begin();it2 != class_freq_map.end(); it2++)
{
total += it2->second;
}
for(std::tr1::unordered_map<string, int>::iterator it3=class_freq_map.begin(); it3 != class_freq_map.end(); it3++)
{
prob = (double) it3->second / total;
entropy -= ( prob * log2(prob));
}
return entropy;
}
void WRclass(enum SC c)
{
dbg_printf("%11s ",str_class(c));
}