void testNormalize() {
BOOST_LOG(rdInfoLog) << "-----------------------\n test normalize"
<< std::endl;
Normalizer normalizer;
// Test sulfoxide normalization.
std::string smi1 = "CS(C)=O";
std::shared_ptr<ROMol> m1(SmilesToMol(smi1));
ROMOL_SPTR normalized(normalizer.normalize(*m1));
TEST_ASSERT(MolToSmiles(*normalized) == "C[S+](C)[O-]");
// normalize sulfone.
std::string smi2 = "C[S+2]([O-])([O-])C";
std::shared_ptr<ROMol> m2(SmilesToMol(smi2));
ROMOL_SPTR normalized2(normalizer.normalize(*m2));
TEST_ASSERT(MolToSmiles(*normalized2) == "CS(C)(=O)=O");
BOOST_LOG(rdInfoLog) << "Finished" << std::endl;
}
int main()
{
cout << ">> TEST DATA SET (k=3)" << endl;
rowvec inX = randu<rowvec>(4);
mat dataSet = randu<mat>(4, 4);
ucolvec labels;
labels << 1 << endr << 3 << endr << 2 << endr << 3 << endr;
inX.print("inX");
dataSet.print("dataSet:");
labels.print("labels:");
cout << "answer: " << knn(inX, dataSet, labels, 3) << endl << endl;
cout << ">> RAND DATA SET (k=10)" << endl;
srand ( time(NULL) );
inX = randu<rowvec>(100);
dataSet = randu<mat>(1000000, 100);
labels = arma::conv_to<ucolvec>::from(4 * randu<colvec>(1000000));
cout << "0s: " << count(labels.begin(), labels.end(), 0) << endl;
cout << "1s: " << count(labels.begin(), labels.end(), 1) << endl;
cout << "2s: " << count(labels.begin(), labels.end(), 2) << endl;
cout << "3s: " << count(labels.begin(), labels.end(), 3) << endl;
cout << "answer: " << knn(inX, dataSet, labels, 10) << endl << endl;
cout << ">> FILE-BASED DATA SET (k=10)" << endl;
mat data;
data.load("datingTestSet2.txt");
dataSet = data.submat(0, 0, data.n_rows-1, data.n_cols-2);
Normalizer <double> normalizer;
dataSet = normalizer.normalize(dataSet);
labels = arma::conv_to<ucolvec>::from(data.submat(0, data.n_cols-1, data.n_rows-1, data.n_cols-1));
rowvec inX1, inX2, inX3;
inX1 << 58732 << 2.454285 << 0.222380 << endr;
inX2 << 6121 << 8.339588 << 1.443357 << endr;
inX3 << 36800 << 12.45 << 0.64 << endr;
cout << "example from the 1st class:" << inX1 << "answer: " << knn(normalizer.normalize(inX1), dataSet, labels, 10) << endl << endl;
cout << "example from the 2nd class:" << inX2 << "answer: " << knn(normalizer.normalize(inX2), dataSet, labels, 10) << endl << endl;
cout << "example from the 3rd class:" << inX3 << "answer: " << knn(normalizer.normalize(inX3), dataSet, labels, 10) << endl << endl;
return 0;
}
