bool
LogisticRegression::solve(double tolerance)
{
if (numObservations() <= 0)
{
THEA_WARNING << "LogisticRegression: Solving empty problem";
has_solution = true;
solution.resize(0);
}
else
{
StdLinearSolver llsq(StdLinearSolver::Method::DEFAULT, StdLinearSolver::Constraint::UNCONSTRAINED);
llsq.setTolerance(tolerance);
typedef Eigen::Map< MatrixX<double, MatrixLayout::ROW_MAJOR> > M;
M a(&llsq_coeffs[0], numObservations(), ndims);
has_solution = llsq.solve(MatrixWrapper<M>(&a), &llsq_consts[0]);
if (has_solution)
solution = Eigen::Map<VectorXd>(const_cast<double *>(llsq.getSolution()), ndims);
else
solution.resize(0);
}
return has_solution;
}
void Interpolate1D::setObservation(const float & x,
const float & y,
const int & idx)
{
if(idx >= numObservations() ) {
addObservation(x, y);
return;
}
m_observations[idx] = Float2(x, y);
}
bool Interpolate1D::learn()
{
const int dim = numObservations();
if(dim<2) {
std::cout<<"Interpolate1D has too few observations "<<dim;
return false;
}
m_xTrain->resize(dim, 1);
m_yTrain->resize(dim, 1);
for(int i=0;i<dim;++i) {
m_xTrain->column(0)[i] = m_observations[i].x;
m_yTrain->column(0)[i] = m_observations[i].y;
}
center_data(*m_xTrain, 1, (float)dim, *m_xMean);
center_data(*m_yTrain, 1, (float)dim, *m_yMean);
if(m_rbf) delete m_rbf;
m_rbf = new RbfKernel<float> (.125f * (m_bound.y - m_bound.x) );
return m_covTrain->create(*m_xTrain, *m_rbf);
}
