sgpp::datadriven::RegressionLearner Class Reference

The RegressionLearner class Solves a regression problem with continuous target vector. More...

#include <RegressionLearner.hpp>

Classes
class	Solver

Public Member Functions
sgpp::base::Grid &	getGrid ()
	getGrid
std::shared_ptr< sgpp::base::Grid >	getGridPtr ()
	getGridPtr
size_t	getGridSize () const
	getGridSize
double	getMSE (sgpp::base::DataMatrix &data, const sgpp::base::DataVector &y)
	getMSE
sgpp::base::DataVector	getWeights () const
	getWeights
sgpp::base::DataVector	predict (sgpp::base::DataMatrix &data)
	predict
	RegressionLearner (sgpp::base::RegularGridConfiguration gridConfig, sgpp::base::AdaptivityConfiguration adaptivityConfig, sgpp::solver::SLESolverConfiguration solverConfig, sgpp::solver::SLESolverConfiguration finalSolverConfig, datadriven::RegularizationConfiguration regularizationConfig)
	RegressionLearner.
	RegressionLearner (sgpp::base::RegularGridConfiguration gridConfig, sgpp::base::AdaptivityConfiguration adaptivityConfig, sgpp::solver::SLESolverConfiguration solverConfig, sgpp::solver::SLESolverConfiguration finalSolverConfig, datadriven::RegularizationConfiguration regularizationConfig, std::set< std::set< size_t > > terms)
	RegressionLearner.
void	setWeights (sgpp::base::DataVector weights)
	setWeights
void	train (sgpp::base::DataMatrix &trainDataset, sgpp::base::DataVector &classes)
	train fits a sparse grid regression model.

Detailed Description

The RegressionLearner class Solves a regression problem with continuous target vector.

Constructor & Destructor Documentation

RegressionLearner() [1/2]

sgpp::datadriven::RegressionLearner::RegressionLearner	(	sgpp::base::RegularGridConfiguration	gridConfig,
		sgpp::base::AdaptivityConfiguration	adaptivityConfig,
		sgpp::solver::SLESolverConfiguration	solverConfig,
		sgpp::solver::SLESolverConfiguration	finalSolverConfig,
		datadriven::RegularizationConfiguration	regularizationConfig,
		std::set< std::set< size_t > >	terms
	)

RegressionLearner.

Parameters

gridConfig
adaptivityConfig
solverConfig	is the solver used during each adaptivity step
finalSolverConfig	is the solver used to build the final model
regularizationConfig
terms	is a vector that contains all desired interaction terms. For example, if we want to include grid points that model an interaction between the first and the second predictor, we would include the vector [1,2] in terms.

RegressionLearner() [2/2]

sgpp::datadriven::RegressionLearner::RegressionLearner	(	sgpp::base::RegularGridConfiguration	gridConfig,
		sgpp::base::AdaptivityConfiguration	adaptivityConfig,
		sgpp::solver::SLESolverConfiguration	solverConfig,
		sgpp::solver::SLESolverConfiguration	finalSolverConfig,
		datadriven::RegularizationConfiguration	regularizationConfig
	)

RegressionLearner.

Parameters

gridConfig
adaptivityConfig
solverConfig	is the solver used during each adaptivity step
finalSolverConfig	is the solver used to build the final model
regularizationConfig

Member Function Documentation

getGrid()

base::Grid & sgpp::datadriven::RegressionLearner::getGrid

(

)

getGrid

Returns

the grid

Referenced by python.uq.analysis.asgc.ASGCKnowledge.ASGCKnowledge::getSparseGridFunction(), python.uq.learner.SimulationLearner.SimulationLearner::learnData(), python.uq.learner.SimulationLearner.SimulationLearner::learnDataWithFolding(), python.uq.learner.SimulationLearner.SimulationLearner::learnDataWithTest(), and python.uq.learner.SimulationLearner.SimulationLearner::refineGrid().

getGridPtr()

std::shared_ptr< sgpp::base::Grid > sgpp::datadriven::RegressionLearner::getGridPtr

(

)

getGridPtr

Returns

shared pointer to the grid

getGridSize()

size_t sgpp::datadriven::RegressionLearner::getGridSize

(

)

const

getGridSize

Returns

the size of the grid

Referenced by python.learner.TrainingStopPolicy.TrainingStopPolicy::hasGridSizeChanged(), and python.learner.TrainingStopPolicy.TrainingStopPolicy::isTrainingComplete().

getMSE()

double sgpp::datadriven::RegressionLearner::getMSE	(	sgpp::base::DataMatrix &	data,
		const sgpp::base::DataVector &	y
	)

getMSE

Parameters

data	is the design matrix
y	is the target

Returns

the mean-squared-error of the prediction of the model for the matrix data

References getMSE(), and predict().

Referenced by getMSE().

getWeights()

base::DataVector sgpp::datadriven::RegressionLearner::getWeights

(

)

const

getWeights

Returns

the weights

predict()

base::DataVector sgpp::datadriven::RegressionLearner::predict

(

sgpp::base::DataMatrix &

data

)

predict

Parameters

data	are observations

Returns

the predicted target for matrix data

References sgpp::op_factory::createOperationMultipleEval().

Referenced by getMSE().

setWeights()

void sgpp::datadriven::RegressionLearner::setWeights

(

sgpp::base::DataVector

weights

)

setWeights

Parameters

weights

are the new weights.

train()

void sgpp::datadriven::RegressionLearner::train	(	sgpp::base::DataMatrix &	trainDataset,
		sgpp::base::DataVector &	classes
	)

train fits a sparse grid regression model.

Parameters

trainDataset	is the design matrix
classes	is the (continuous) target

References sgpp::datadriven::RegressionLearner::Solver::cg, sgpp::op_factory::createOperationMultipleEval(), sgpp::base::DataMatrix::getNrows(), sgpp::base::DataVector::getSize(), and sgpp::base::AdaptivityConfiguration::numRefinements_.

---

The documentation for this class was generated from the following files:

• datadriven/src/sgpp/datadriven/application/RegressionLearner.hpp
• datadriven/src/sgpp/datadriven/application/RegressionLearner.cpp