Detailed Description

A class to perform 1D regressions.

It works on a time serie and uses either ad-hoc methods or matrix arithmetic to perform an arbitrary fit. Currently, the following models are supported:

Specific fits:
Least Square fits:

The various variogram models can be found in "Statistics for spatial data", Noel A. C. Cressie, John Wiley & Sons, revised edition, 1993, pp63.

Author: Mathias Bavay

Date: 2011-01-20

#include <libfit1D.h>

Public Types
enum	regression { ZERO , SIMPLE_LINEAR , NOISY_LINEAR , LINVARIO , EXPVARIO , SPHERICVARIO , RATQUADVARIO , LINEARLS , QUADRATIC , POLYNOMIAL }
	Keywords for regression model. More...

Public Member Functions
	Fit1D ()
	Empty Constructor. The model must be set afterwards. If the model has not been set before calling other methods, a nullptr pointer exception will be thrown. More...

	Fit1D (const regression &regType, const std::vector< double > &in_X, const std::vector< double > &in_Y, const bool &updatefit=true)
	Constructor. More...

	Fit1D (const std::string &regType, const std::vector< double > &in_X, const std::vector< double > &in_Y, const bool &updatefit=true)
	Constructor for user provided model type. More...

	Fit1D (const Fit1D &i_fit)
	Copy constructor. More...

	~Fit1D ()

bool	setModel (const regression &i_regType, const std::vector< double > &in_X, const std::vector< double > &in_Y, const bool &updatefit=true)
	Set or reset the regression model. More...

bool	setModel (const std::string &i_regType, const std::vector< double > &in_X, const std::vector< double > &in_Y, const bool &updatefit=true)
	Set or reset the regression model. More...

void	setGuess (const std::vector< double > &lambda_in)
	Provide a set of initial values for the model parameters. The model can be used right after providing the guesses, and it would use those guesses as parameters, thus allowing the user to force his model parameters. More...

void	setLapseRate (const double &lapse_rate)
	Set a forced lapse rate for linear regressions This will throw an exception for all other regression models! More...

void	setDegree (const size_t &degree)
	Set the degree of the polynomial regression This will throw an exception for all other regression models! More...

bool	fit ()
	Compute the regression parameters. More...

double	f (const double &x) const
	Calculate a value using the computed least square fit. The fit has to be computed before. More...

std::vector< double >	getParams () const
	Calculate the parameters of the fit. The fit has to be computed before. More...

std::string	getName () const
	Return the name of the fit model. More...

std::string	getInfo () const
	Return a string of information about the fit. The fit has to be computed before. More...

void	setInfo (const std::string &info)
	Set the information string. This is useful to append some extra information to the information string. This should be called after computing the fit (otherwise it will be overwritten). More...

Fit1D &	operator= (const Fit1D &source)

double	operator() (const double &x) const
	Calculate a value using the computed least square fit. The fit has to be computed before. More...

std::string	toString () const

Member Enumeration Documentation

◆ regression

enum mio::Fit1D::regression

Keywords for regression model.

Enumerator
ZERO	always return zero (this is a way to disable detrending)
SIMPLE_LINEAR	basic, cheap linear fit
NOISY_LINEAR	same as SIMPLE_LINEAR but trying to remove outliers
LINVARIO	linear variogram
EXPVARIO	exponential variogram
SPHERICVARIO	spherical variogram
RATQUADVARIO	rational quadratic variogram
LINEARLS	linear, using least squares
QUADRATIC	quadratic
POLYNOMIAL	polynomial regression

Constructor & Destructor Documentation

◆ Fit1D() [1/4]

mio::Fit1D::Fit1D ( )

inline

Empty Constructor. The model must be set afterwards. If the model has not been set before calling other methods, a nullptr pointer exception will be thrown.

◆ Fit1D() [2/4]

mio::Fit1D::Fit1D	(	const regression &	regType,
		const std::vector< double > &	in_X,
		const std::vector< double > &	in_Y,
		const bool &	updatefit = `true`
	)

Constructor.

Parameters

regType	regression model to use
in_X	vector of data points abscissae
in_Y	vector of data points ordinates
updatefit	should the fit be redone? (default=true, otherwise you must manually call fit())

◆ Fit1D() [3/4]

mio::Fit1D::Fit1D	(	const std::string &	regType,
		const std::vector< double > &	in_X,
		const std::vector< double > &	in_Y,
		const bool &	updatefit = `true`
	)

Constructor for user provided model type.

Parameters

regType	regression model to use
in_X	vector of data points abscissae
in_Y	vector of data points ordinates
updatefit	should the fit be redone? (default=true, otherwise you must manually call fit())

◆ Fit1D() [4/4]

mio::Fit1D::Fit1D ( const Fit1D & i_fit )

Copy constructor.

Parameters

i_fit Object to copy

◆ ~Fit1D()

mio::Fit1D::~Fit1D ( )

inline

Member Function Documentation

◆ f()

double mio::Fit1D::f ( const double & x ) const

inline

Calculate a value using the computed least square fit. The fit has to be computed before.

Parameters

x abscissa

Returns: f(x) using the computed least square fit

◆ fit()

bool mio::Fit1D::fit ( )

inline

Compute the regression parameters.

This is a specific optimizations for 1D linear regression: y = ax + b with scalars. See https://mathworld.wolfram.com/LeastSquaresFitting.html

Returns: false if could not compute the parameters

◆ getInfo()

std::string mio::Fit1D::getInfo ( ) const

inline

Return a string of information about the fit. The fit has to be computed before.

Returns: info string

◆ getName()

std::string mio::Fit1D::getName ( ) const

inline

Return the name of the fit model.

Returns: model name

◆ getParams()

std::vector< double > mio::Fit1D::getParams ( ) const

inline

Calculate the parameters of the fit. The fit has to be computed before.

Returns: vector containing the coefficients

◆ operator()()

double mio::Fit1D::operator() ( const double & x ) const

inline

Calculate a value using the computed least square fit. The fit has to be computed before.

Parameters

x abscissa

Returns: f(x) using the computed least square fit

◆ operator=()

Fit1D & mio::Fit1D::operator= ( const Fit1D & source )

◆ setDegree()

void mio::Fit1D::setDegree ( const size_t & degree )

inline

Set the degree of the polynomial regression This will throw an exception for all other regression models!

Parameters

degree degree of the polynomial regression to set

◆ setGuess()

void mio::Fit1D::setGuess ( const std::vector< double > & lambda_in )

inline

Provide a set of initial values for the model parameters. The model can be used right after providing the guesses, and it would use those guesses as parameters, thus allowing the user to force his model parameters.

Parameters

lambda_in one initial value per model parameter

◆ setInfo()

void mio::Fit1D::setInfo ( const std::string & info )

inline

Set the information string. This is useful to append some extra information to the information string. This should be called after computing the fit (otherwise it will be overwritten).

Parameters

info string

◆ setLapseRate()

void mio::Fit1D::setLapseRate ( const double & lapse_rate )

inline

Set a forced lapse rate for linear regressions This will throw an exception for all other regression models!

Parameters

lapse_rate lapse rate to set

◆ setModel() [1/2]

bool mio::Fit1D::setModel	(	const regression &	i_regType,
		const std::vector< double > &	in_X,
		const std::vector< double > &	in_Y,
		const bool &	updatefit = `true`
	)

Set or reset the regression model.

Parameters

i_regType	regression model to use
in_X	vector of data points abscissae
in_Y	vector of data points ordinates
updatefit	should the fit be redone? (default=true, otherwise you must manually call fit())

Returns: false if could not compute the parameters. if !updatefit, always return true

◆ setModel() [2/2]

bool mio::Fit1D::setModel	(	const std::string &	i_regType,
		const std::vector< double > &	in_X,
		const std::vector< double > &	in_Y,
		const bool &	updatefit = `true`
	)

Set or reset the regression model.

Parameters

i_regType	regression model to use
in_X	vector of data points abscissae
in_Y	vector of data points ordinates
updatefit	should the fit be redone? (default=true, otherwise you must manually call fit())

Returns: false if could not compute the parameters. if !updatefit, always return true

◆ toString()

std::string mio::Fit1D::toString ( ) const

inline

The documentation for this class was generated from the following files:

Detailed Description

Public Types

Public Member Functions

Member Enumeration Documentation

◆ regression

Constructor & Destructor Documentation

◆ Fit1D() [1/4]

◆ Fit1D() [2/4]

◆ Fit1D() [3/4]

◆ Fit1D() [4/4]

◆ ~Fit1D()

Member Function Documentation

◆ f()

◆ fit()

◆ getInfo()

◆ getName()

◆ getParams()

◆ operator()()

◆ operator=()

◆ setDegree()

◆ setGuess()

◆ setInfo()

◆ setLapseRate()

◆ setModel() [1/2]

◆ setModel() [2/2]

◆ toString()