StabilisedGLLS Class Reference

Numerically stabilised general linear least squares. More...

#include <qle/math/stabilisedglls.hpp>

Public Types
enum	Method { None , MaxAbs , MeanStdDev }

Public Member Functions
template<class xContainer, class yContainer, class vContainer>
	StabilisedGLLS (const xContainer &x, const yContainer &y, const vContainer &v, const Method method=MeanStdDev)
const Array &	transformedCoefficients () const
const Array &	transformedResiduals () const
const Array &	transformedStandardErrors () const
const Array &	transformedError () const
const Array &	xMultiplier () const
	Transformation parameters (u => (u + shift) * multiplier for u = x, y).
const Array &	xShift () const
const Real	yMultiplier () const
const Real	yShift () const
Size	size () const
Size	dim () const
template<class xType, class vContainer>
Real	eval (xType x, vContainer &v, typename boost::enable_if< typename boost::is_arithmetic< xType >::type >::type *=0)
	evaluate regression function in terms of original x, y
template<class xType, class vContainer>
Real	eval (xType x, vContainer &v, typename boost::disable_if< typename boost::is_arithmetic< xType >::type >::type *=0)
	evaluate regression function in terms of original x, y

Protected Member Functions
template<class xContainer, class yContainer, class vContainer>
void	calculate (xContainer x, yContainer y, vContainer v, typename boost::enable_if< typename boost::is_arithmetic< typename xContainer::value_type >::type >::type *=0)
template<class xContainer, class yContainer, class vContainer>
void	calculate (xContainer x, yContainer y, vContainer v, typename boost::disable_if< typename boost::is_arithmetic< typename xContainer::value_type >::type >::type *=0)

Protected Attributes
Array	a_
Array	err_
Array	residuals_
Array	standardErrors_
Array	xMultiplier_
Array	xShift_
Real	yMultiplier_
Real	yShift_
Method	method_
QuantLib::ext::shared_ptr< GeneralLinearLeastSquares >	glls_

Detailed Description

Numerically stabilised general linear least squares.

The input data is linearly transformed before performing the linear least squares fit. The linear least squares fit on the transformed data is done using the GeneralLinearLeastSquares class.