LeastSquaresMetaModelSelection

class LeastSquaresMetaModelSelection(*args)

Least squares metamodel selection factory.

Adaptative sparse selection, as proposed in [blatman2009].

Methods

getBasisSequenceFactory()	Accessor to the basis sequence factory.
getClassName()	Accessor to the object's name.
getCoefficients()	Accessor to the coefficients.
getFittingAlgorithm()	Accessor to the fitting algorithm.
getName()	Accessor to the object's name.
getPsi()	Accessor to the basis.
getRelativeError()	Accessor to the coefficients.
getResidual()	Accessor to the coefficients.
getWeight()	Accessor to the weights.
getX()	Accessor to the input sample.
getY()	Accessor to the output sample.
hasName()	Test if the object is named.
involvesModelSelection()	Get the model selection flag.
run(*args)	Run the algorithm.
setBasisSequenceFactory(basisSequenceFactory)	Set the basis sequence factory.
setFittingAlgorithm(fittingAlgorithm)	Set the fitting algorithm.
setName(name)	Accessor to the object's name.

See also

ApproximationAlgorithm, PenalizedLeastSquaresAlgorithm

Notes

The LeastSquaresMetaModelSelection is built from a LeastSquaresMetaModelSelectionFactory. The stopping criteria for the model selection is defined through the following entries of the ResourceMap:

• LeastSquaresMetaModelSelection-ErrorThreshold: if the error computed by cross validation is lesser than this threshold,

  then the exploration is stopped. The default value is 0: this criteria is not activated by default.

• LeastSquaresMetaModelSelection-MaximumError: if the error computed by cross validation is greater than this threshold,

  then the exploration is stopped (the error first decreases then increases when the basis complexity increases). The best approximation obtained so far is returned. The default value is 0.5.

• LeastSquaresMetaModelSelection-alpha: through the exploration, the minimum error is stored. If the

  current error is greater than , then the exploration is stopped. The best approximation obtained so far is returned. The default value is 2. This value allows one to filter little fluctuations in the error computation.

__init__(*args)

getBasisSequenceFactory()

Accessor to the basis sequence factory.

Returns:

basisSequenceFactoryBasisSequenceFactory

Basis sequence factory.

getClassName()

Accessor to the object’s name.

Returns:

class_namestr

The object class name (object.__class__.__name__).

getCoefficients()

Accessor to the coefficients.

Returns:

coefficientsPoint

The coefficients

getFittingAlgorithm()

Accessor to the fitting algorithm.

Returns:

fittingAlgorithmFittingAlgorithm

Fitting algorithm.

getName()

Accessor to the object’s name.

Returns:

namestr

The name of the object.

getPsi()

Accessor to the basis.

Returns:

coefficientsBasis

The basis

getRelativeError()

Accessor to the coefficients.

Returns:

relativeErrorfloat

The relative error

getResidual()

Accessor to the coefficients.

Returns:

coefficientsfloat

The residual

getWeight()

Accessor to the weights.

Returns:

weightPoint

Output weights

getX()

Accessor to the input sample.

Returns:

xSample

Input sample

getY()

Accessor to the output sample.

Returns:

ySample

Input sample

hasName()

Test if the object is named.

Returns:

hasNamebool

True if the name is not empty.

involvesModelSelection()

Get the model selection flag.

A model selection method can be used to select the coefficients of the decomposition which enable to best predict the output. Model selection can lead to a sparse functional chaos expansion.

Returns:

involvesModelSelectionbool

True if the method involves a model selection method.

run(*args)

Run the algorithm.

setBasisSequenceFactory(basisSequenceFactory)

Set the basis sequence factory.

Parameters:

basisSequenceFactoryBasisSequenceFactory

Basis sequence factory.

setFittingAlgorithm(fittingAlgorithm)

Set the fitting algorithm.

Parameters:

fittingAlgorithmFittingAlgorithm

Fitting algorithm.

setName(name)

Accessor to the object’s name.

Parameters:

namestr

The name of the object.