Package weka.classifiers.meta

Class MultiClassClassifier

  • All Implemented Interfaces:
    java.io.Serializable, java.lang.Cloneable, Classifier, BatchPredictor, CapabilitiesHandler, CapabilitiesIgnorer, CommandlineRunnable, OptionHandler, Randomizable, RevisionHandler, WeightedInstancesHandler
    Direct Known Subclasses:
    MultiClassClassifierUpdateable
    public class MultiClassClassifier
extends RandomizableSingleClassifierEnhancer
implements OptionHandler, WeightedInstancesHandler
    A metaclassifier for handling multi-class datasets with 2-class classifiers. This classifier is also capable of applying error correcting output codes for increased accuracy. If the base classifier cannot handle instance weights, and the instance weights are not uniform, the data will be resampled with replacement based on the weights before being passed to the base classifier.

    Valid options are: 

     -M <num>
  Sets the method to use. Valid values are 0 (1-against-all),
  1 (random codes), 2 (exhaustive code), and 3 (1-against-1). (default 0)
 
     -R <num>
  Sets the multiplier when using random codes. (default 2.0)
     -P
  Use pairwise coupling (only has an effect for 1-against1)
     -L
  Use log loss decoding for random and exhaustive codes.
     -S <num>
  Random number seed.
  (default 1)
     -D
  If set, classifier is run in debug mode and
  may output additional info to the console
     -W
  Full name of base classifier.
  (default: weka.classifiers.functions.Logistic)
     
 Options specific to classifier weka.classifiers.functions.Logistic:
 
     -D
  Turn on debugging output.
     -R <ridge>
  Set the ridge in the log-likelihood.
     -M <number>
  Set the maximum number of iterations (default -1, until convergence).
    Version:
    $Revision: 15476 $
    Author:
    Eibe Frank (eibe@cs.waikato.ac.nz), Len Trigg (len@reeltwo.com), Richard Kirkby (rkirkby@cs.waikato.ac.nz)
    See Also:
    Serialized Form

    • Field Detail

      • METHOD_1_AGAINST_ALL

        public static final int METHOD_1_AGAINST_ALL
        1-against-all
        See Also:
        Constant Field Values

      • METHOD_ERROR_RANDOM

        public static final int METHOD_ERROR_RANDOM
        random correction code
        See Also:
        Constant Field Values

      • METHOD_ERROR_EXHAUSTIVE

        public static final int METHOD_ERROR_EXHAUSTIVE
        exhaustive correction code
        See Also:
        Constant Field Values

      • METHOD_1_AGAINST_1

        public static final int METHOD_1_AGAINST_1
        1-against-1
        See Also:
        Constant Field Values

      • TAGS_METHOD

        public static final Tag[] TAGS_METHOD
        The error correction modes

    • Constructor Detail

      • MultiClassClassifier

        public MultiClassClassifier()
        Constructor.

    • Method Detail

      • buildClassifier

        public void buildClassifier​(Instances insts)
                     throws java.lang.Exception
        Builds the classifiers.
        Specified by:
        buildClassifier in interface Classifier
        Parameters:
        insts - the training data.
        Throws:
        java.lang.Exception - if a classifier can't be built

      • individualPredictions

        public double[] individualPredictions​(Instance inst)
                               throws java.lang.Exception
        Returns the individual predictions of the base classifiers for an instance. Used by StackedMultiClassClassifier. Returns the probability for the second "class" predicted by each base classifier.
        Parameters:
        inst - the instance to get the prediction for
        Returns:
        the individual predictions
        Throws:
        java.lang.Exception - if the predictions can't be computed successfully

      • distributionForInstance

        public double[] distributionForInstance​(Instance inst)
                                 throws java.lang.Exception
        Returns the distribution for an instance.
        Specified by:
        distributionForInstance in interface Classifier
        Overrides:
        distributionForInstance in class AbstractClassifier
        Parameters:
        inst - the instance to get the distribution for
        Returns:
        the distribution
        Throws:
        java.lang.Exception - if the distribution can't be computed successfully

      • toString

        public java.lang.String toString()
        Prints the classifiers.
        Overrides:
        toString in class java.lang.Object
        Returns:
        a string representation of the classifier

      • setOptions

        public void setOptions​(java.lang.String[] options)
                throws java.lang.Exception
        Parses a given list of options.

        Valid options are: 

         -M <num>
  Sets the method to use. Valid values are 0 (1-against-all),
  1 (random codes), 2 (exhaustive code), and 3 (1-against-1). (default 0)
 
         -R <num>
  Sets the multiplier when using random codes. (default 2.0)
         -P
  Use pairwise coupling (only has an effect for 1-against1)
         -L
  Use log loss decoding for random and exhaustive codes.
         -S <num>
  Random number seed.
  (default 1)
         -D
  If set, classifier is run in debug mode and
  may output additional info to the console
         -W
  Full name of base classifier.
  (default: weka.classifiers.functions.Logistic)
         
 Options specific to classifier weka.classifiers.functions.Logistic:
 
         -D
  Turn on debugging output.
         -R <ridge>
  Set the ridge in the log-likelihood.
         -M <number>
  Set the maximum number of iterations (default -1, until convergence).
        Specified by:
        setOptions in interface OptionHandler
        Overrides:
        setOptions in class RandomizableSingleClassifierEnhancer
        Parameters:
        options - the list of options as an array of strings
        Throws:
        java.lang.Exception - if an option is not supported

      • globalInfo

        public java.lang.String globalInfo()
        Returns:
        a description of the classifier suitable for displaying in the explorer/experimenter gui

      • logLossDecodingTipText

        public java.lang.String logLossDecodingTipText()
        Returns:
        tip text for this property suitable for displaying in the explorer/experimenter gui

      • getLogLossDecoding

        public boolean getLogLossDecoding()
        Whether log loss decoding is used for random or exhaustive codes.
        Returns:
        true if log loss is used

      • setLogLossDecoding

        public void setLogLossDecoding​(boolean newlogLossDecoding)
        Sets whether log loss decoding is used for random or exhaustive codes.
        Parameters:
        newlogLossDecoding - true if log loss is to be used

      • randomWidthFactorTipText

        public java.lang.String randomWidthFactorTipText()
        Returns:
        tip text for this property suitable for displaying in the explorer/experimenter gui

      • getRandomWidthFactor

        public double getRandomWidthFactor()
        Gets the multiplier when generating random codes. Will generate numClasses * m_RandomWidthFactor codes.
        Returns:
        the width multiplier

      • setRandomWidthFactor

        public void setRandomWidthFactor​(double newRandomWidthFactor)
        Sets the multiplier when generating random codes. Will generate numClasses * m_RandomWidthFactor codes.
        Parameters:
        newRandomWidthFactor - the new width multiplier

      • methodTipText

        public java.lang.String methodTipText()
        Returns:
        tip text for this property suitable for displaying in the explorer/experimenter gui

      • getMethod

        public SelectedTag getMethod()
        Gets the method used. Will be one of METHOD_1_AGAINST_ALL, METHOD_ERROR_RANDOM, METHOD_ERROR_EXHAUSTIVE, or METHOD_1_AGAINST_1.
        Returns:
        the current method.

      • setMethod

        public void setMethod​(SelectedTag newMethod)
        Sets the method used. Will be one of METHOD_1_AGAINST_ALL, METHOD_ERROR_RANDOM, METHOD_ERROR_EXHAUSTIVE, or METHOD_1_AGAINST_1.
        Parameters:
        newMethod - the new method.

      • setUsePairwiseCoupling

        public void setUsePairwiseCoupling​(boolean p)
        Set whether to use pairwise coupling with 1-vs-1 classification to improve probability estimates.
        Parameters:
        p - true if pairwise coupling is to be used

      • getUsePairwiseCoupling

        public boolean getUsePairwiseCoupling()
        Gets whether to use pairwise coupling with 1-vs-1 classification to improve probability estimates.
        Returns:
        true if pairwise coupling is to be used

      • usePairwiseCouplingTipText

        public java.lang.String usePairwiseCouplingTipText()
        Returns:
        tip text for this property suitable for displaying in the explorer/experimenter gui

      • pairwiseCoupling

        public static double[] pairwiseCoupling​(double[][] n,
                                        double[][] r)
        Implements pairwise coupling.
        Parameters:
        n - the sum of weights used to train each model
        r - the probability estimate from each model
        Returns:
        the coupled estimates

      • main

        public static void main​(java.lang.String[] argv)
        Main method for testing this class.
        Parameters:
        argv - the options