CurveFitterLibraries
Classes and methods for fitting growth curves
 All Classes Namespaces Files Functions Variables Enumerations Enumerator Properties Events
Public Types | Protected Member Functions | Protected Attributes | List of all members
GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian Class Referenceabstract

A wrapper class for the alglib functions, used to fit non-linear models More...

Inheritance diagram for GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian:
GrowthCurveLibrary.AbstractFitter GrowthCurveLibrary.IAbstractFitter GrowthCurveLibrary.ExponentialFit

Public Types

enum  TerminationType : int {
  NotYetSet = -2, WrongParams = -1, SumOfSquaresChangeConverge = 1, ParameterConverge = 2,
  MaxIterations = 5
}
 

Protected Member Functions

abstract void EvaluateSumofSquaresFunction (double[] Params, ref double func, object obj)
 Used by the fitting routine to evaluate possible values, meant to stay compatible with alg lig More...
 
abstract void EvaluateSumofSquaresFunction22 (double[] Params, double[] func, object obj)
 
abstract void EvaluateSumOfSquaresGradient (double[] Params, ref double func, double[] GradientVec, object obj)
 
abstract void EvaluateSumOfSquaresGradient22 (double[] Params, double[] fi, double[,] GradientVec, object obj)
 
abstract void CreateHessianDelegates ()
 
abstract double[] CreateInitialParameterGuess ()
 
alglib.minlmreport GeneralLMFitting ()
 See the example in alglib on levenberg marquadt to understand this code. Based entirely on their example, somethings kept around for compatibility More...
 
- Protected Member Functions inherited from GrowthCurveLibrary.AbstractFitter
void VerifyInput (double[] XDATA, double[] YDATA)
 
void makeYHAT ()
 
double calculateRMSE ()
 
double calculateR2 ()
 
abstract void FitModel ()
 

Protected Attributes

Delegates.DerivativePortionCalculator[,] HessianFunctions
 
double EPSX
 stopping criterion. Algorithm stops if |X(k+1)-X(k)| <= EpsX*(1+|X(k)|) More...
 
- Protected Attributes inherited from GrowthCurveLibrary.AbstractFitter
double[] x
 
double[] y
 
double[] ypred
 
double[] pParameters
 

Additional Inherited Members

- Public Member Functions inherited from GrowthCurveLibrary.AbstractFitter
double CalculateLogLikelihoodAssumingNormallyDistributedError (IEnumerable< double > xPoints, IEnumerable< double > yPoints)
 
double[] CalculateResidualsAtNewPoints (double[] xVals, double[] yvals)
 
 AbstractFitter ()
 
double calculateResidualSumofSquares ()
 Only should be called by external classes following a fit More...
 
virtual double calculateAbsError ()
 
abstract double FunctiontoFit (double x)
 
virtual void GenerateFitLine (double LowX, double interval, double HighX, out double[] xvalues, out double[] yvalues)
 
virtual IEnumerable< double > MakePredictionsAtPoints (IEnumerable< double > XValues)
 
- Public Attributes inherited from GrowthCurveLibrary.AbstractFitter
string name = ""
 
- Properties inherited from GrowthCurveLibrary.AbstractFitter
string Comment [get, set]
 
bool FitByDefault [get, set]
 
int NumberOfParameters [get]
 
double[] X [get]
 
double[] Y [get]
 
double[] PredictedValues [get]
 
double[] Parameters [get]
 
bool SuccessfulFit [get, set]
 
double RMSE [get]
 
double R2 [get]
 
double AbsError [get]
 
double[] Residuals [get]
 

Detailed Description

A wrapper class for the alglib functions, used to fit non-linear models

Poorly named class, see fitting method for comments as to why, this no longer uses hessian

Member Enumeration Documentation

enum GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian.TerminationType : int
Enumerator
NotYetSet 
WrongParams 
SumOfSquaresChangeConverge 
ParameterConverge 
MaxIterations 

Member Function Documentation

abstract void GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian.CreateHessianDelegates ( )
protectedpure virtual

Implemented in GrowthCurveLibrary.ExponentialFit.

abstract double [] GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian.CreateInitialParameterGuess ( )
protectedpure virtual

Implemented in GrowthCurveLibrary.ExponentialFit.

abstract void GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian.EvaluateSumofSquaresFunction ( double[]  Params,
ref double  func,
object  obj 
)
protectedpure virtual

Used by the fitting routine to evaluate possible values, meant to stay compatible with alg lig

Parameters
Params
Returns

Implemented in GrowthCurveLibrary.ExponentialFit.

abstract void GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian.EvaluateSumofSquaresFunction22 ( double[]  Params,
double[]  func,
object  obj 
)
protectedpure virtual

Implemented in GrowthCurveLibrary.ExponentialFit.

abstract void GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian.EvaluateSumOfSquaresGradient ( double[]  Params,
ref double  func,
double[]  GradientVec,
object  obj 
)
protectedpure virtual

Implemented in GrowthCurveLibrary.ExponentialFit.

abstract void GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian.EvaluateSumOfSquaresGradient22 ( double[]  Params,
double[]  fi,
double  GradientVec[,],
object  obj 
)
protectedpure virtual

Implemented in GrowthCurveLibrary.ExponentialFit.

alglib.minlmreport GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian.GeneralLMFitting ( )
protected

See the example in alglib on levenberg marquadt to understand this code. Based entirely on their example, somethings kept around for compatibility

Returns

Member Data Documentation

double GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian.EPSX
protected

stopping criterion. Algorithm stops if |X(k+1)-X(k)| <= EpsX*(1+|X(k)|)

Delegates.DerivativePortionCalculator [,] GrowthCurveLibrary.NonLinearFitterWithGradientAndHessian.HessianFunctions
protected
The documentation for this class was generated from the following file: