Implements a fitter for the Gumbel distribution. More...

#include <OpenMS/MATH/STATISTICS/GumbelMaxLikelihoodFitter.h>

Classes
struct	GumbelDistributionFitResult
	struct to represent the parameters of a gumbel distribution More...

Public Member Functions
	GumbelMaxLikelihoodFitter ()
	Default constructor.

	GumbelMaxLikelihoodFitter (GumbelDistributionFitResult init)
	Default constructor.

virtual	~GumbelMaxLikelihoodFitter ()
	Destructor.

void	setInitialParameters (const GumbelDistributionFitResult &result)
	sets the gumbel distribution start parameters a and b for the fitting

GumbelDistributionFitResult	fitWeighted (const std::vector< double > &x, const std::vector< double > &w)
	Fits a Gumbel (maximum) distribution to the data points `x` with weights `w` by weighted maximum likelihood.

Protected Attributes
GumbelDistributionFitResult	init_param_

Private Member Functions
	GumbelMaxLikelihoodFitter (const GumbelMaxLikelihoodFitter &rhs)
	Copy constructor (not implemented)

GumbelMaxLikelihoodFitter &	operator= (const GumbelMaxLikelihoodFitter &rhs)
	assignment operator (not implemented)

Detailed Description

Implements a fitter for the Gumbel distribution.

This class fits a Gumbel distribution to a number of data points. The results as well as the initial guess are specified using the struct GumbelDistributionFitResult.

The formula with the fitted parameters can be transformed into a gnuplot formula using getGnuplotFormula() after fitting.

Constructor & Destructor Documentation

◆ GumbelMaxLikelihoodFitter() [1/3]

GumbelMaxLikelihoodFitter ( )

Default constructor.

◆ GumbelMaxLikelihoodFitter() [2/3]

GumbelMaxLikelihoodFitter ( GumbelDistributionFitResult init )

Default constructor.

◆ ~GumbelMaxLikelihoodFitter()

virtual ~GumbelMaxLikelihoodFitter ( )

virtual

Destructor.

◆ GumbelMaxLikelihoodFitter() [3/3]

GumbelMaxLikelihoodFitter ( const GumbelMaxLikelihoodFitter & rhs )

private

Copy constructor (not implemented)

Member Function Documentation

◆ fitWeighted()

GumbelDistributionFitResult fitWeighted	(	const std::vector< double > &	x,
		const std::vector< double > &	w
	)

Fits a Gumbel (maximum) distribution to the data points x with weights w by weighted maximum likelihood.

The scale parameter is the unique root of the (monotone) profile score equation, which is solved by safeguarded Newton iterations; the location parameter then follows in closed form. Data points with non-finite values or non-positive weights are ignored. The scale parameter of the initial parameters is the starting point of the iterations (warm start). If no data point remains, the initial parameters are returned. The fitted parameters are used as initial parameters of the next call.

Parameters

[in]	x	Input x values
[in]	w	Input weights (same length as `x`)

Exceptions

Exception::InvalidParameter	is thrown if `x` and `w` differ in size
Exception::UnableToFit	is thrown if fitting cannot be performed (e.g. the data have no spread)

◆ operator=()

GumbelMaxLikelihoodFitter & operator= ( const GumbelMaxLikelihoodFitter & rhs )

private

assignment operator (not implemented)

◆ setInitialParameters()

void setInitialParameters ( const GumbelDistributionFitResult & result )

sets the gumbel distribution start parameters a and b for the fitting

Member Data Documentation

◆ init_param_

GumbelDistributionFitResult init_param_

protected

Classes

Public Member Functions

Protected Attributes

Private Member Functions