RegOptimizer

java.lang.Object
- keel.Algorithms.SVM.SMO.supportVector.RegOptimizer

All Implemented Interfaces:

java.io.Serializable, OptionHandler

Direct Known Subclasses:

RegSMO
```
public class RegOptimizer
extends java.lang.Object
implements OptionHandler, java.io.Serializable
```
Base class implementation for learning algorithm of SVMreg Valid options are:
```
 -L <double>
  The epsilon parameter in epsilon-insensitive loss function.
  (default 1.0e-3)
```
```
 -W <double>
  The random number seed.
  (default 1)
```
Version:

$Revision: 1.1 $

Author:

Remco Bouckaert (remco@cs.waikato.ac.nz,rrb@xm.co.nz)

See Also:

Serialized Form

Field Summary

Fields
Modifier and Type	Field and Description
`double[]`	`m_alpha` alpha and alpha* arrays containing weights for solving dual problem
`double[]`	`m_alphaStar` alpha and alpha* arrays containing weights for solving dual problem
`protected double`	`m_b` offset
`protected boolean`	`m_bModelBuilt` flag to indicate whether the model is built yet
`protected double`	`m_C` capacity parameter, copied from SVMreg
`protected int`	`m_classIndex` index of class variable in data set
`protected Instances`	`m_data` points to data set
`protected double`	`m_epsilon` epsilon of epsilon-insensitive cost function
`protected Kernel`	`m_kernel` the kernel
`protected int`	`m_nCacheHits` number of kernel cache hits, used for printing statistics only
`protected int`	`m_nEvals` number of kernel evaluations, used for printing statistics only
`protected int`	`m_nInstances` number of instances in data set
`protected int`	`m_nSeed` seed for initializing random number generator
`protected java.util.Random`	`m_random` random number generator
`protected int[]`	`m_sparseIndices` Variables to hold indices vector in sparse form.
`protected double[]`	`m_sparseWeights` Variables to hold weight vector in sparse form.
`protected SMOset`	`m_supportVectors` set of support vectors, that is, vectors with alpha(*)!
`protected SVMreg`	`m_SVM` parent SVMreg class
`protected double[]`	`m_target` class values/desired output vector
`protected double[]`	`m_weights` weights for linear kernel

Constructor Summary

Constructors
Constructor and Description

RegOptimizer()
the default constructor

Constructors
Constructor and Description
`RegOptimizer()` the default constructor

Method Summary

All Methods Instance Methods Concrete Methods
Modifier and Type	Method and Description
`void`	`buildClassifier(Instances data)` learn SVM parameters from data.
`java.lang.String`	`epsilonParameterTipText()` Returns the tip text for this property
`int`	`getCacheHits()` return the number of kernel cache hits
`double`	`getEpsilonParameter()` Get the value of epsilon parameter of the epsilon insensitive loss function.
`int`	`getKernelEvaluations()` returns the number of kernel evaluations
`java.lang.String[]`	`getOptions()` Gets the current settings of the classifier.
`protected double`	`getScore()` Compute the value of the objective function.
`int`	`getSeed()` Gets the current seed value for the random number generator
`protected void`	`init(Instances data)` initializes the algorithm
`java.util.Enumeration`	`listOptions()` Gets an enumeration describing the available options.
`boolean`	`modelBuilt()` flag to indicate whether the model was built yet
`java.lang.String`	`seedTipText()` Returns the tip text for this property
`void`	`setEpsilonParameter(double v)` Set the value of epsilon parameter of the epsilon insensitive loss function.
`void`	`setOptions(java.lang.String[] options)` Parses a given list of options.
`void`	`setSeed(int value)` Sets the seed value for the random number generator
`void`	`setSVMReg(SVMreg value)` sets the parent SVM
`double`	`SVMOutput(Instance inst)`
`protected double`	`SVMOutput(int index)` SVMOutput of an instance in the training set, m_data This uses the cache, unlike SVMOutput(Instance)
`protected void`	`wrapUp()` wrap up various variables to save memeory and do some housekeeping after optimization has finished.

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

- Field Detail
  - m_alpha
```
public double[] m_alpha
```
    alpha and alpha* arrays containing weights for solving dual problem
  - m_alphaStar
```
public double[] m_alphaStar
```
    alpha and alpha* arrays containing weights for solving dual problem
  - m_b
```
protected double m_b
```
    offset
  - m_epsilon
```
protected double m_epsilon
```
    epsilon of epsilon-insensitive cost function
  - m_C
```
protected double m_C
```
    capacity parameter, copied from SVMreg
  - m_target
```
protected double[] m_target
```
    class values/desired output vector
  - m_data
```
protected Instances m_data
```
    points to data set
  - m_kernel
```
protected Kernel m_kernel
```
    the kernel
  - m_classIndex
```
protected int m_classIndex
```
    index of class variable in data set
  - m_nInstances
```
protected int m_nInstances
```
    number of instances in data set
  - m_random
```
protected java.util.Random m_random
```
    random number generator
  - m_nSeed
```
protected int m_nSeed
```
    seed for initializing random number generator
  - m_supportVectors
```
protected SMOset m_supportVectors
```
    set of support vectors, that is, vectors with alpha(*)!=0
  - m_nEvals
```
protected int m_nEvals
```
    number of kernel evaluations, used for printing statistics only
  - m_nCacheHits
```
protected int m_nCacheHits
```
    number of kernel cache hits, used for printing statistics only
  - m_weights
```
protected double[] m_weights
```
    weights for linear kernel
  - m_sparseWeights
```
protected double[] m_sparseWeights
```
    Variables to hold weight vector in sparse form. (To reduce storage requirements.)
  - m_sparseIndices
```
protected int[] m_sparseIndices
```
    Variables to hold indices vector in sparse form. (To reduce storage requirements.)
  - m_bModelBuilt
```
protected boolean m_bModelBuilt
```
    flag to indicate whether the model is built yet
  - m_SVM
```
protected SVMreg m_SVM
```
    parent SVMreg class
- Constructor Detail
  - RegOptimizer
```
public RegOptimizer()
```
    the default constructor
- Method Detail
  - listOptions
```
public java.util.Enumeration listOptions()
```
    Gets an enumeration describing the available options.
    
    Specified by:
    
    listOptions in interface OptionHandler
    
    Returns:
    
    an enumeration of all the available options.
  - setOptions
```
public void setOptions(java.lang.String[] options)
                throws java.lang.Exception
```
    Parses a given list of options.
    Valid options are:
```
 -L <double>
  The epsilon parameter in epsilon-insensitive loss function.
  (default 1.0e-3)
```
```
 -W <double>
  The random number seed.
  (default 1)
```
    Specified by:
    
    setOptions in interface OptionHandler
    
    Parameters:
    
    options - the list of options as an array of strings
    
    Throws:
    
    java.lang.Exception - if an option is not supported
  - getOptions
```
public java.lang.String[] getOptions()
```
    Gets the current settings of the classifier.
    
    Specified by:
    
    getOptions in interface OptionHandler
    
    Returns:
    
    an array of strings suitable for passing to setOptions
  - modelBuilt
```
public boolean modelBuilt()
```
    flag to indicate whether the model was built yet
    
    Returns:
    
    true if the model was built
  - setSVMReg
```
public void setSVMReg(SVMreg value)
```
    sets the parent SVM
    
    Parameters:
    
    value - the parent SVM
  - getKernelEvaluations
```
public int getKernelEvaluations()
```
    returns the number of kernel evaluations
    
    Returns:
    
    the number of kernel evaluations
  - getCacheHits
```
public int getCacheHits()
```
    return the number of kernel cache hits
    
    Returns:
    
    the number of hits
  - init
```
protected void init(Instances data)
             throws java.lang.Exception
```
    initializes the algorithm
    
    Parameters:
    
    data - the data to work with
    
    Throws:
    
    java.lang.Exception - if m_SVM is null
  - wrapUp
```
protected void wrapUp()
               throws java.lang.Exception
```
    wrap up various variables to save memeory and do some housekeeping after optimization has finished.
    
    Throws:
    
    java.lang.Exception - if something goes wrong
  - getScore
```
protected double getScore()
                   throws java.lang.Exception
```
    Compute the value of the objective function.
    
    Returns:
    
    the score
    
    Throws:
    
    java.lang.Exception - if something goes wrong
  - buildClassifier
```
public void buildClassifier(Instances data)
                     throws java.lang.Exception
```
    learn SVM parameters from data. Subclasses should implement something more interesting.
    
    Parameters:
    
    data - the data to work with
    
    Throws:
    
    java.lang.Exception - always an Exceoption since subclasses must override it
  - SVMOutput
```
protected double SVMOutput(int index)
                    throws java.lang.Exception
```
    SVMOutput of an instance in the training set, m_data This uses the cache, unlike SVMOutput(Instance)
    
    Parameters:
    
    index - index of the training instance in m_data
    
    Returns:
    
    the SVM output
    
    Throws:
    
    java.lang.Exception - if something goes wrong
  - SVMOutput
```
public double SVMOutput(Instance inst)
                 throws java.lang.Exception
```
    Parameters:
    
    inst -
    
    Returns:
    
    Throws:
    
    java.lang.Exception
  - seedTipText
```
public java.lang.String seedTipText()
```
    Returns the tip text for this property
    
    Returns:
    
    tip text for this property suitable for displaying in the explorer/experimenter gui
  - getSeed
```
public int getSeed()
```
    Gets the current seed value for the random number generator
    
    Returns:
    
    the seed value
  - setSeed
```
public void setSeed(int value)
```
    Sets the seed value for the random number generator
    
    Parameters:
    
    value - the seed value
  - epsilonParameterTipText
```
public java.lang.String epsilonParameterTipText()
```
    Returns the tip text for this property
    
    Returns:
    
    tip text for this property suitable for displaying in the explorer/experimenter gui
  - getEpsilonParameter
```
public double getEpsilonParameter()
```
    Get the value of epsilon parameter of the epsilon insensitive loss function.
    
    Returns:
    
    Value of epsilon parameter.
  - setEpsilonParameter
```
public void setEpsilonParameter(double v)
```
    Set the value of epsilon parameter of the epsilon insensitive loss function.
    
    Parameters:
    
    v - Value to assign to epsilon parameter.

Class RegOptimizer

Field Summary

Constructor Summary

Method Summary

Methods inherited from class java.lang.Object

Field Detail

m_alpha

m_alphaStar

m_b

m_epsilon

m_C

m_target

m_data

m_kernel

m_classIndex

m_nInstances

m_random

m_nSeed

m_supportVectors

m_nEvals

m_nCacheHits

m_weights

m_sparseWeights

m_sparseIndices

m_bModelBuilt

m_SVM

Constructor Detail

RegOptimizer

Method Detail

listOptions

setOptions

getOptions

modelBuilt

setSVMReg

getKernelEvaluations

getCacheHits

init

wrapUp

getScore

buildClassifier

SVMOutput

SVMOutput

seedTipText

getSeed

setSeed

epsilonParameterTipText

getEpsilonParameter

setEpsilonParameter