SVDRegressionCalculation Class

Class SVDRegressionCalculation computes linear regression parameters by the method of least squares using a singular value decomposition.

Inheritance Hierarchy

SystemObject
CenterSpace.NMath.CoreSVDRegressionCalculation

Namespace: CenterSpace.NMath.Core
Assembly: NMath (in NMath.dll) Version: 7.3

Syntax

[SerializableAttribute]
public class SVDRegressionCalculation : IRegressionCalculation, 
	ICloneable

<SerializableAttribute>
Public Class SVDRegressionCalculation
	Implements IRegressionCalculation, ICloneable

[SerializableAttribute]
public ref class SVDRegressionCalculation : IRegressionCalculation, 
	ICloneable

[<SerializableAttribute>]
type SVDRegressionCalculation =  
    class
        interface IRegressionCalculation
        interface ICloneable
    end

The SVDRegressionCalculation type exposes the following members.

Constructors

	Name	Description
	SVDRegressionCalculation	Constructs an SVDRegressionCalculation instance with all sizes equal to zero.
	SVDRegressionCalculation(DoubleMatrix)	Constructs an SVDRegressionCalculation instance from the given matrix.
	SVDRegressionCalculation(DoubleMatrix, Double)	Constructs a SVDRegressionCalculation instance from the given matrix. The specified tolerance is used in computing the numerical rank of the matrix.

Properties

	Name	Description
	Cols	Gets the number of columns in the matrix.
	Fail	Gets the status of the singular value decomposition.
	IsGood	Returns true if the singular value decomposition may be used to solve least squares problems; otherwise false.
	Rank	Gets the numerical rank of the matrix.
	RankAvailable	Returns the rank if it was calculated as a byproduct of the parameter calculation.
	Rows	Gets the number of rows in the matrix.
	Tolerance	Gets and sets the tolerance for computing the numerical rank and SVD tuncation. In truncation all singular values less than Tolerance are set to zero and solutions will be computed using the truncated SVD. If the Tolerance is set to 0 no truncation will be performed. The default tolerance is 0, i.e. no truncation.
	XTXInv	Gets the matrix formed by taking the inverse of the product of the transpose of the regression matrix with itself, if available.
	XTXInvAvailable	Gets a boolean indicating whether or not the matrix formed by taking the inverse of the product of the transpose of the regression matrix with itself is avaialble as part of the decomposition.

Methods

	Name	Description
	CalculateParameters(DoubleMatrix, DoubleVector)	Calculates the parameters for the regression using a singular value decomposition of the regression matrix to solve the least squares problem.
	CalculateParameters(DoubleMatrix, DoubleVector, Boolean)	Calculates the parameters for the regression using a singular value decomposition of the regression matrix to solve the least squares problem.
	Clone	Creates a deep copy of this regression calculator instance.
	Factor(DoubleMatrix)	Factors a given matrix so that it may be used to solve least squares problems.
	Factor(DoubleMatrix, Double)	Factors a given matrix so that it may be used to solve least squares problems. The specified tolerance is used in computing the numerical rank of the matrix.
	OnSerializing	Checks for soln_ to be null, instantiates if so
	Solve	Computes the solution to the least squares problem Ax = b.

Remarks

Class SVDRegressionCalculation finds the minimal norm solution to the overdetermined linear system:

Ax = b

That is, this class finds the vector x that minimizes the 2-norm of the residual vector Ax - b.