Namespaces

Types in MathNet.Numerics.LinearRegression

Type MultipleRegression

Namespace MathNet.Numerics.LinearRegression

Static Functions

Public Static Functions

Vector<T> DirectMethod<T>(Matrix<T> x, Vector<T> y, DirectRegressionMethod method)

Find the model parameters β such that X*β with predictor X becomes as close to response Y as possible, with least squares residuals.
Parameters
Matrix<T> x

Predictor matrix X

Vector<T> y

Response vector Y

DirectRegressionMethod method

The direct method to be used to compute the regression.

Return
Vector<T>

Best fitting vector for model parameters β

Matrix<T> DirectMethod<T>(Matrix<T> x, Matrix<T> y, DirectRegressionMethod method)

Find the model parameters β such that X*β with predictor X becomes as close to response Y as possible, with least squares residuals.
Parameters
Matrix<T> x

Predictor matrix X

Matrix<T> y

Response matrix Y

DirectRegressionMethod method

The direct method to be used to compute the regression.

Return
Matrix<T>

Best fitting vector for model parameters β

T[] DirectMethod<T>(T[][] x, T[] y, bool intercept, DirectRegressionMethod method)

T[] DirectMethod<T>(IEnumerable<Tuple<T[], T>> samples, bool intercept, DirectRegressionMethod method)

Find the model parameters β such that their linear combination with all predictor-arrays in X become as close to their response in Y as possible, with least squares residuals. Uses the cholesky decomposition of the normal equations.
Parameters
IEnumerable<Tuple<T[], T>> samples

Sequence of predictor-arrays and their response.

bool intercept

True if an intercept should be added as first artificial predictor value. Default = false.

DirectRegressionMethod method

The direct method to be used to compute the regression.

Return
T[]

Best fitting list of model parameters β for each element in the predictor-arrays.

Vector<T> NormalEquations<T>(Matrix<T> x, Vector<T> y)

Find the model parameters β such that X*β with predictor X becomes as close to response Y as possible, with least squares residuals. Uses the cholesky decomposition of the normal equations.
Parameters
Matrix<T> x

Predictor matrix X

Vector<T> y

Response vector Y

Return
Vector<T>

Best fitting vector for model parameters β

Matrix<T> NormalEquations<T>(Matrix<T> x, Matrix<T> y)

Find the model parameters β such that X*β with predictor X becomes as close to response Y as possible, with least squares residuals. Uses the cholesky decomposition of the normal equations.
Parameters
Matrix<T> x

Predictor matrix X

Matrix<T> y

Response matrix Y

Return
Matrix<T>

Best fitting vector for model parameters β

T[] NormalEquations<T>(T[][] x, T[] y, bool intercept)

T[] NormalEquations<T>(IEnumerable<Tuple<T[], T>> samples, bool intercept)

Find the model parameters β such that their linear combination with all predictor-arrays in X become as close to their response in Y as possible, with least squares residuals. Uses the cholesky decomposition of the normal equations.
Parameters
IEnumerable<Tuple<T[], T>> samples

Sequence of predictor-arrays and their response.

bool intercept

True if an intercept should be added as first artificial predictor value. Default = false.

Return
T[]

Best fitting list of model parameters β for each element in the predictor-arrays.

Matrix<T> QR<T>(Matrix<T> x, Matrix<T> y)

Find the model parameters β such that X*β with predictor X becomes as close to response Y as possible, with least squares residuals. Uses an orthogonal decomposition and is therefore more numerically stable than the normal equations but also slower.
Parameters
Matrix<T> x

Predictor matrix X

Matrix<T> y

Response matrix Y

Return
Matrix<T>

Best fitting vector for model parameters β

T[] QR<T>(T[][] x, T[] y, bool intercept)

Vector<T> QR<T>(Matrix<T> x, Vector<T> y)

Find the model parameters β such that X*β with predictor X becomes as close to response Y as possible, with least squares residuals. Uses an orthogonal decomposition and is therefore more numerically stable than the normal equations but also slower.
Parameters
Matrix<T> x

Predictor matrix X

Vector<T> y

Response vector Y

Return
Vector<T>

Best fitting vector for model parameters β

T[] QR<T>(IEnumerable<Tuple<T[], T>> samples, bool intercept)

Find the model parameters β such that their linear combination with all predictor-arrays in X become as close to their response in Y as possible, with least squares residuals. Uses an orthogonal decomposition and is therefore more numerically stable than the normal equations but also slower.
Parameters
IEnumerable<Tuple<T[], T>> samples

Sequence of predictor-arrays and their response.

bool intercept

True if an intercept should be added as first artificial predictor value. Default = false.

Return
T[]

Best fitting list of model parameters β for each element in the predictor-arrays.

Vector<T> Svd<T>(Matrix<T> x, Vector<T> y)

Find the model parameters β such that X*β with predictor X becomes as close to response Y as possible, with least squares residuals. Uses a singular value decomposition and is therefore more numerically stable (especially if ill-conditioned) than the normal equations or QR but also slower.
Parameters
Matrix<T> x

Predictor matrix X

Vector<T> y

Response vector Y

Return
Vector<T>

Best fitting vector for model parameters β

Matrix<T> Svd<T>(Matrix<T> x, Matrix<T> y)

Find the model parameters β such that X*β with predictor X becomes as close to response Y as possible, with least squares residuals. Uses a singular value decomposition and is therefore more numerically stable (especially if ill-conditioned) than the normal equations or QR but also slower.
Parameters
Matrix<T> x

Predictor matrix X

Matrix<T> y

Response matrix Y

Return
Matrix<T>

Best fitting vector for model parameters β

T[] Svd<T>(T[][] x, T[] y, bool intercept)

T[] Svd<T>(IEnumerable<Tuple<T[], T>> samples, bool intercept)

Find the model parameters β such that their linear combination with all predictor-arrays in X become as close to their response in Y as possible, with least squares residuals. Uses a singular value decomposition and is therefore more numerically stable (especially if ill-conditioned) than the normal equations or QR but also slower.
Parameters
IEnumerable<Tuple<T[], T>> samples

Sequence of predictor-arrays and their response.

bool intercept

True if an intercept should be added as first artificial predictor value. Default = false.

Return
T[]

Best fitting list of model parameters β for each element in the predictor-arrays.