yat
0.20.3pre
|
#include <yat/regression/Multivariate.h>
Public Member Functions | |
virtual | ~Multivariate (void) |
destructor | |
virtual void | fit (const utility::Matrix &x, const utility::VectorBase &y)=0 |
virtual void | fit2 (const utility::MatrixBase &x, const utility::VectorBase &y) |
virtual const utility::Vector & | fit_parameters (void) const =0 |
virtual double | predict (const utility::VectorBase &x) const =0 |
predict value in x according to model | |
The name of this class is a bit a misnomer; a better name would be Multivariable. It defines the interface for multivariable regression classes, i.e., is a vector in the general model .
|
pure virtual |
This is the old interface before MatrixBase was introduced and is only kept not to break compatibility with older code. This function should have the same behaviour as the fit2 function and typically it's implemented with a direct call two fit2.
Implemented in theplu::yat::regression::NegativeBinomial, theplu::yat::regression::MultiDimensional, and theplu::yat::regression::Poisson.
|
virtual |
Estimating model parameters based on X to fit output data y. Each row in X corresponds to one data point, i.e., number of rows in X must match size of y.
A default implementation (relying on function fit()) is provided just to not break older code. The implementation involves a copying of the input matrix and classes inheriting from Multivariate should this preferably implement this function to avoid that copying.
Reimplemented in theplu::yat::regression::NegativeBinomial, theplu::yat::regression::MultiDimensional, and theplu::yat::regression::Poisson.
|
pure virtual |
Implemented in theplu::yat::regression::NegativeBinomial, theplu::yat::regression::MultiDimensional, and theplu::yat::regression::Poisson.