void BipartiteGraph::eraseEdge( size_t n1, size_t n2 ) {
DAI_ASSERT( n1 < nrNodes1() );
DAI_ASSERT( n2 < nrNodes2() );
size_t iter;
// Search for edge among neighbors of n1
for( iter = 0; iter < nb1(n1).size(); iter++ )
if( nb1(n1, iter).node == n2 ) {
// Remove it
nb1(n1).erase( nb1(n1).begin() + iter );
break;
}
// Change the iter and dual values of the subsequent neighbors
for( ; iter < nb1(n1).size(); iter++ ) {
Neighbor &m2 = nb1( n1, iter );
m2.iter = iter;
nb2( m2.node, m2.dual ).dual = iter;
}
// Search for edge among neighbors of n2
for( iter = 0; iter < nb2(n2).size(); iter++ )
if( nb2(n2, iter).node == n1 ) {
// Remove it
nb2(n2).erase( nb2(n2).begin() + iter );
break;
}
// Change the iter and node values of the subsequent neighbors
for( ; iter < nb2(n2).size(); iter++ ) {
Neighbor &m1 = nb2( n2, iter );
m1.iter = iter;
nb1( m1.node, m1.dual ).dual = iter;
}
}
double nball(double *x, double *par)
{
// if (x[0]>=0.0 && x[0]<1.0) return nb0( par[0], par[1], 0.0, (1-par[4])*ntotal*(1-par[2]), (1-par[4])*ntotal*par[2], par[4]*ntotal*(1-par[2]), par[4]*ntotal*par[2], 2, par[3]);
// else if (x[0]>=1.0 && x[0]<2.0) return nb1( par[0], par[1], 0.0, (1-par[4])*ntotal*(1-par[2]), (1-par[4])*ntotal*par[2], par[4]*ntotal*(1-par[2]), par[4]*ntotal*par[2], 2, par[3]);
// else if (x[0]>=2.0 && x[0]<3.0) return nb2( par[0], par[1], 0.0, (1-par[4])*ntotal*(1-par[2]), (1-par[4])*ntotal*par[2], par[4]*ntotal*(1-par[2]), par[4]*ntotal*par[2], 2, par[3]);
// else if (x[0]>=3.0 && x[0]<4.0) return nb3( par[0], par[1], 0.0, (1-par[4])*ntotal*(1-par[2]), (1-par[4])*ntotal*par[2], par[4]*ntotal*(1-par[2]), par[4]*ntotal*par[2], 2, par[3]);
if (x[0]>=0.0 && x[0]<1.0) return nb0( par[0], par[1], 0.0, (1-par[4])*ntotal*(1-par[2]), (1-par[4])*ntotal*par[2], par[4]*ntotal*0.14, par[4]*ntotal*0.80, 0, par[4]*ntotal*0.03, 2, par[3]);
else if (x[0]>=1.0 && x[0]<2.0) return nb1( par[0], par[1], 0.0, (1-par[4])*ntotal*(1-par[2]), (1-par[4])*ntotal*par[2], par[4]*ntotal*0.14, par[4]*ntotal*0.80, 0, par[4]*ntotal*0.03, 2, par[3]);
else if (x[0]>=2.0 && x[0]<3.0) return nb2( par[0], par[1], 0.0, (1-par[4])*ntotal*(1-par[2]), (1-par[4])*ntotal*par[2], par[4]*ntotal*0.14, par[4]*ntotal*0.80, 0, par[4]*ntotal*0.03, 2, par[3]);
else if (x[0]>=3.0 && x[0]<4.0) return nb3( par[0], par[1], 0.0, (1-par[4])*ntotal*(1-par[2]), (1-par[4])*ntotal*par[2], par[4]*ntotal*0.14, par[4]*ntotal, 0, par[4]*ntotal*0.03, 2, par[3]);
else if (x[0]>=4.0 && x[0]<5.0) return nb4( par[0], par[1], 0.0, (1-par[4])*ntotal*(1-par[2]), (1-par[4])*ntotal*par[2], par[4]*ntotal*0.14, par[4]*ntotal, 0, par[4]*ntotal*0.03, 2, par[3]);
}
void BipartiteGraph::eraseNode1( size_t n1 ) {
DAI_ASSERT( n1 < nrNodes1() );
// Erase neighbor entry of node n1
_nb1.erase( _nb1.begin() + n1 );
// Adjust neighbor entries of nodes of type 2
for( size_t n2 = 0; n2 < nrNodes2(); n2++ ) {
for( size_t iter = 0; iter < nb2(n2).size(); ) {
Neighbor &m1 = nb2(n2, iter);
if( m1.node == n1 ) {
// delete this entry, because it points to the deleted node
nb2(n2).erase( nb2(n2).begin() + iter );
} else if( m1.node > n1 ) {
// update this entry and the corresponding dual of the neighboring node of type 1
m1.iter = iter;
m1.node--;
nb1( m1.node, m1.dual ).dual = iter;
iter++;
} else {
// skip
iter++;
}
}
}
}
void BipartiteGraph::eraseNode2( size_t n2 ) {
DAI_ASSERT( n2 < nrNodes2() );
// Erase neighbor entry of node n2
_nb2.erase( _nb2.begin() + n2 );
// Adjust neighbor entries of nodes of type 1
for( size_t n1 = 0; n1 < nrNodes1(); n1++ ) {
for( size_t iter = 0; iter < nb1(n1).size(); ) {
Neighbor &m2 = nb1(n1, iter);
if( m2.node == n2 ) {
// delete this entry, because it points to the deleted node
nb1(n1).erase( nb1(n1).begin() + iter );
} else if( m2.node > n2 ) {
// update this entry and the corresponding dual of the neighboring node of type 2
m2.iter = iter;
m2.node--;
nb2( m2.node, m2.dual ).dual = iter;
iter++;
} else {
// skip
iter++;
}
}
}
}
int main()
{
std::ofstream out("prediction.txt");
Dataset dataMy;
Dataset dataMisha;
Dataset dataTeach;
Dataset dataTest, dataTest1;
std::string str = "train1.csv";
std::string str1 = "test1.csv";
LoadTeachDataset(str, &dataTeach, 2);
LoadTestDataset(str1, &dataTest, 2);
LoadTestDataset(str1, &dataTest1, 2);
vector<double> h, h1;
// дл¤ ¤дра ≈пачечникова
// h.push_back(3.111);
// h.push_back(2.349);
// дл¤ квартического ¤дра
// h.push_back(3.83);
// h.push_back(3.2);
h.push_back(3.9);
h.push_back(3.1);
NaiveBayesKernelOtherDensity nb2(2, h);
nb2.Learn(dataTeach);
nb2.Classify(&dataTest);
// NaiveBayesKernelOtherDensity nb22(2, h1);
// nb22.Learn(dataTeach);
// nb22.Classify(&dataTest1);
//size_t err = LOO_h_vector<NaiveBayesKernelOtherDensity>(h, dataTeach);
//std::cout << err << std::endl;
// double minH = searchHvector_Force<NaiveBayesKernelOtherDensity>(2.1, 4.1, 0.1, h, 1, dataTeach);
// std::cout << minH << std::endl;
for (size_t index = 0; index < dataTest.size(); ++index)
{
out << dataTest[index].class_label << std::endl;
dataMy.push_back(dataTest[index]);
//dataMisha.push_back(dataTest1[index]);
}
Dataset _dataTeach;
Dataset _dataTest, _dataTest1;
std::string _str = "train2.csv";
std::string _str1 = "test2.csv";
LoadTeachDataset(_str, &_dataTeach, 10);
LoadTestDataset(_str1, &_dataTest, 10);
LoadTestDataset(_str1, &_dataTest1, 10);
vector<double> _h, _h1;
// дл¤ ¤дра ≈пачечникова
// _h.push_back(10);
// _h.push_back(10.8);
// _h.push_back(0.2);
// _h.push_back(0.156);
// _h.push_back(0.1325);
// _h.push_back(171);
// _h.push_back(89.7);
// _h.push_back(52.4);
// _h.push_back(16.8);
// _h.push_back(81.3);
// дл¤ квартического ¤дра
// _h.push_back(10.3102);
// _h.push_back(10.1);
// _h.push_back(0.2);
// _h.push_back(0.157);
// _h.push_back(0.1325);
// _h.push_back(171);
// _h.push_back(89.6);
// _h.push_back(53.1);
// _h.push_back(16.8);
// _h.push_back(81);
_h.push_back(10.3112);
_h.push_back(10.1);
_h.push_back(0.2);
_h.push_back(0.157);
_h.push_back(0.1335);
_h.push_back(169);
_h.push_back(90.5);
_h.push_back(53.1);
_h.push_back(16.8);
_h.push_back(81.4);
// size_t _err = LOO_h_vector<NaiveBayesKernelOtherDensity>(_h, _dataTeach);
//std::cout << _err << std::endl;
// double _minH = searchHvector_Force<NaiveBayesKernelOtherDensity>(80.4, 82.4, 0.1, _h, 9, _dataTeach);
// std::cout << _minH << std::endl;
NaiveBayesKernelOtherDensity _nb(2, _h);
_nb.Learn(_dataTeach);
_nb.Classify(&_dataTest);
// NaiveBayesKernelOtherDensity _nb2(2, _h1);
// _nb2.Learn(_dataTeach);
// _nb2.Classify(&_dataTest1);
for (size_t index = 0; index < _dataTest.size(); ++index)
{
out << _dataTest[index].class_label << std::endl;
dataMy.push_back(_dataTest[index]);
//dataMisha.push_back(_dataTest1[index]);
}
out.close();
// vector<vector<size_t>> result(2, vector<size_t>(2));
// size_t res = DataTest(dataMisha, dataMy, result);
// std::cout << (double)res << std::endl;
//std::ifstream fin("out1.csv");
std::ifstream fin("ans.txt");
if (!fin)
{
std::cout << "err";
return 0;
}
std::cout << "begin" << std::endl;
while (!fin.eof())
{
size_t class_label;
fin >> class_label;
Object obj;
obj.class_label = class_label;
dataMisha.push_back(obj);
}
vector<vector<size_t>> result(2, vector<size_t>(2));
size_t res = DataTest(dataMisha, dataMy, result);
std::cout << (double)res / 8340.0 << std::endl;
return 0;
}
