double ConsistencySubsetEvaluator::evaluateSubset(const std::vector<int>& columns,
TgsProgress*)
{
BinHashFunctor::_columns = columns;
const DataFrame& df = _dataFrame;
// hash<bin key, hash<class enumeration, instance count> >
typedef HashMap<
std::vector<int>, // bin key
InconsistentInstancesMap, // map of class enumerations to instance count
BinHashFunctor> // fancy hash function and comparison that only looks at 'columns'
BinMap;
BinMap binMap;
for (unsigned int i = 0; i < _bins.size(); i++)
{
//BinMap::iterator it = binMap.find(_bins[i]);
InconsistentInstancesMap& eim = binMap[_bins[i]];
int classEnum = _enumMap[df.getTrainingLabel(i)];
if (eim.size() == 0)
{
eim.resize(_enumMap.size(), 0);
}
eim[classEnum]++;
}
int inconsistencyCount = 0;
for (BinMap::const_iterator it = binMap.begin(); it != binMap.end(); it++)
{
inconsistencyCount += _calculateInconsistentCount(it->second);
}
double inconsistencyRate = (double)inconsistencyCount / (double)_bins.size();
return -inconsistencyRate;
}
inline void dump(std::ostream& s) const
{
BinMap::const_iterator bmi;
for (bmi = bins.begin(); bmi != bins.end(); bmi++)
{
const BinRange& br = bmi->first;
s << std::right << std::setw(3) << br.first
<< "-" << std::left << std::setw(3) << br.second
<< ": " << std::right << bmi->second
<< std::endl;
}
s << std::right << std::setw(3) << bins.begin()->first.first
<< "-" << std::left << std::setw(3) << bins.rbegin()->first.second
<< ": " << std::right << total
<< std::endl;
};
inline void addValue(double v)
{
BinMap::iterator bm_itr;
for (bm_itr=bins.begin(); bm_itr != bins.end(); bm_itr++)
{
const BinRange& range = bm_itr->first;
if (range.first < v && v <= range.second)
{
bm_itr->second++;
total++;
break;
}
}
}
