Class RaceSearch

  • All Implemented Interfaces:
    java.io.Serializable, RankedOutputSearch, OptionHandler, RevisionHandler, TechnicalInformationHandler

    public class RaceSearch
    extends ASSearch
    implements RankedOutputSearch, OptionHandler, TechnicalInformationHandler
    Races the cross validation error of competing attribute subsets. Use in conjuction with a ClassifierSubsetEval. RaceSearch has four modes:

    forward selection races all single attribute additions to a base set (initially no attributes), selects the winner to become the new base set and then iterates until there is no improvement over the base set.

    Backward elimination is similar but the initial base set has all attributes included and races all single attribute deletions.

    Schemata search is a bit different. Each iteration a series of races are run in parallel. Each race in a set determines whether a particular attribute should be included or not---ie the race is between the attribute being "in" or "out". The other attributes for this race are included or excluded randomly at each point in the evaluation. As soon as one race has a clear winner (ie it has been decided whether a particular attribute should be inor not) then the next set of races begins, using the result of the winning race from the previous iteration as new base set.

    Rank race first ranks the attributes using an attribute evaluator and then races the ranking. The race includes no attributes, the top ranked attribute, the top two attributes, the top three attributes, etc.

    It is also possible to generate a raked list of attributes through the forward racing process. If generateRanking is set to true then a complete forward race will be run---that is, racing continues until all attributes have been selected. The order that they are added in determines a complete ranking of all the attributes.

    Racing uses paired and unpaired t-tests on cross-validation errors of competing subsets. When there is a significant difference between the means of the errors of two competing subsets then the poorer of the two can be eliminated from the race. Similarly, if there is no significant difference between the mean errors of two competing subsets and they are within some threshold of each other, then one can be eliminated from the race.

    For more information see:

    Andrew W. Moore, Mary S. Lee: Efficient Algorithms for Minimizing Cross Validation Error. In: Eleventh International Conference on Machine Learning, 190-198, 1994.

    BibTeX:

     @inproceedings{Moore1994,
        author = {Andrew W. Moore and Mary S. Lee},
        booktitle = {Eleventh International Conference on Machine Learning},
        pages = {190-198},
        publisher = {Morgan Kaufmann},
        title = {Efficient Algorithms for Minimizing Cross Validation Error},
        year = {1994}
     }
     

    Valid options are:

     -R <0 = forward | 1 = backward race | 2 = schemata | 3 = rank>
      Type of race to perform.
      (default = 0).
     -L <significance>
      Significance level for comaparisons
      (default = 0.001(forward/backward/rank)/0.01(schemata)).
     -T <threshold>
      Threshold for error comparison.
      (default = 0.001).
     -A <attribute evaluator>
      Attribute ranker to use if doing a 
      rank search. Place any
      evaluator options LAST on 
      the command line following a "--".
      eg. -A weka.attributeSelection.GainRatioAttributeEval ... -- -M.
      (default = GainRatioAttributeEval)
     -F <0 = 10 fold | 1 = leave-one-out>
      Folds for cross validation
      (default = 0 (1 if schemata race)
     -Q
      Generate a ranked list of attributes.
      Forces the search to be forward
      and races until all attributes have
      selected, thus producing a ranking.
     -N <num to select>
      Specify number of attributes to retain from 
      the ranking. Overides -T. Use in conjunction with -Q
     -J <threshold>
      Specify a theshold by which attributes
      may be discarded from the ranking.
      Use in conjuction with -Q
     -Z
      Verbose output for monitoring the search.
     
     Options specific to evaluator weka.attributeSelection.GainRatioAttributeEval:
     
     -M
      treat missing values as a seperate value.
    Version:
    $Revision: 1.26 $
    Author:
    Mark Hall (mhall@cs.waikato.ac.nz)
    See Also:
    Serialized Form
    • Field Detail

      • TAGS_SELECTION

        public static final Tag[] TAGS_SELECTION
      • XVALTAGS_SELECTION

        public static final Tag[] XVALTAGS_SELECTION
    • Constructor Detail

      • RaceSearch

        public RaceSearch()
    • Method Detail

      • globalInfo

        public java.lang.String globalInfo()
        Returns a string describing this search method
        Returns:
        a description of the search method suitable for displaying in the explorer/experimenter gui
      • getTechnicalInformation

        public TechnicalInformation getTechnicalInformation()
        Returns an instance of a TechnicalInformation object, containing detailed information about the technical background of this class, e.g., paper reference or book this class is based on.
        Specified by:
        getTechnicalInformation in interface TechnicalInformationHandler
        Returns:
        the technical information about this class
      • raceTypeTipText

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

        public void setRaceType​(SelectedTag d)
        Set the race type
        Parameters:
        d - the type of race
      • getRaceType

        public SelectedTag getRaceType()
        Get the race type
        Returns:
        the type of race
      • significanceLevelTipText

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

        public void setSignificanceLevel​(double sig)
        Sets the significance level to use
        Parameters:
        sig - the significance level
      • getSignificanceLevel

        public double getSignificanceLevel()
        Get the significance level
        Returns:
        the current significance level
      • thresholdTipText

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

        public void setThreshold​(double t)
        Sets the threshold for comparisons
        Specified by:
        setThreshold in interface RankedOutputSearch
        Parameters:
        t - the threshold to use
      • getThreshold

        public double getThreshold()
        Get the threshold
        Specified by:
        getThreshold in interface RankedOutputSearch
        Returns:
        the current threshold
      • foldsTypeTipText

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

        public void setFoldsType​(SelectedTag d)
        Set the xfold type
        Parameters:
        d - the type of xval
      • getFoldsType

        public SelectedTag getFoldsType()
        Get the xfold type
        Returns:
        the type of xval
      • debugTipText

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

        public void setDebug​(boolean d)
        Set whether verbose output should be generated.
        Parameters:
        d - true if output is to be verbose.
      • getDebug

        public boolean getDebug()
        Get whether output is to be verbose
        Returns:
        true if output will be verbose
      • attributeEvaluatorTipText

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

        public void setAttributeEvaluator​(ASEvaluation newEvaluator)
        Set the attribute evaluator to use for generating the ranking.
        Parameters:
        newEvaluator - the attribute evaluator to use.
      • getAttributeEvaluator

        public ASEvaluation getAttributeEvaluator()
        Get the attribute evaluator used to generate the ranking.
        Returns:
        the evaluator used to generate the ranking.
      • generateRankingTipText

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

        public void setGenerateRanking​(boolean doRank)
        Records whether the user has requested a ranked list of attributes.
        Specified by:
        setGenerateRanking in interface RankedOutputSearch
        Parameters:
        doRank - true if ranking is requested
      • getGenerateRanking

        public boolean getGenerateRanking()
        Gets whether ranking has been requested. This is used by the AttributeSelection module to determine if rankedAttributes() should be called.
        Specified by:
        getGenerateRanking in interface RankedOutputSearch
        Returns:
        true if ranking has been requested.
      • numToSelectTipText

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

        public void setNumToSelect​(int n)
        Specify the number of attributes to select from the ranked list (if generating a ranking). -1 indicates that all attributes are to be retained.
        Specified by:
        setNumToSelect in interface RankedOutputSearch
        Parameters:
        n - the number of attributes to retain
      • getNumToSelect

        public int getNumToSelect()
        Gets the number of attributes to be retained.
        Specified by:
        getNumToSelect in interface RankedOutputSearch
        Returns:
        the number of attributes to retain
      • getCalculatedNumToSelect

        public int getCalculatedNumToSelect()
        Gets the calculated number of attributes to retain. This is the actual number of attributes to retain. This is the same as getNumToSelect if the user specifies a number which is not less than zero. Otherwise it should be the number of attributes in the (potentially transformed) data.
        Specified by:
        getCalculatedNumToSelect in interface RankedOutputSearch
      • selectionThresholdTipText

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

        public void setSelectionThreshold​(double threshold)
        Set the threshold by which the AttributeSelection module can discard attributes.
        Parameters:
        threshold - the threshold.
      • getSelectionThreshold

        public double getSelectionThreshold()
        Returns the threshold so that the AttributeSelection module can discard attributes from the ranking.
      • listOptions

        public java.util.Enumeration listOptions()
        Returns 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:

         -R <0 = forward | 1 = backward race | 2 = schemata | 3 = rank>
          Type of race to perform.
          (default = 0).
         -L <significance>
          Significance level for comaparisons
          (default = 0.001(forward/backward/rank)/0.01(schemata)).
         -T <threshold>
          Threshold for error comparison.
          (default = 0.001).
         -A <attribute evaluator>
          Attribute ranker to use if doing a 
          rank search. Place any
          evaluator options LAST on 
          the command line following a "--".
          eg. -A weka.attributeSelection.GainRatioAttributeEval ... -- -M.
          (default = GainRatioAttributeEval)
         -F <0 = 10 fold | 1 = leave-one-out>
          Folds for cross validation
          (default = 0 (1 if schemata race)
         -Q
          Generate a ranked list of attributes.
          Forces the search to be forward
          and races until all attributes have
          selected, thus producing a ranking.
         -N <num to select>
          Specify number of attributes to retain from 
          the ranking. Overides -T. Use in conjunction with -Q
         -J <threshold>
          Specify a theshold by which attributes
          may be discarded from the ranking.
          Use in conjuction with -Q
         -Z
          Verbose output for monitoring the search.
         
         Options specific to evaluator weka.attributeSelection.GainRatioAttributeEval:
         
         -M
          treat missing values as a seperate value.
        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 BestFirst.
        Specified by:
        getOptions in interface OptionHandler
        Returns:
        an array of strings suitable for passing to setOptions()
      • search

        public int[] search​(ASEvaluation ASEval,
                            Instances data)
                     throws java.lang.Exception
        Searches the attribute subset space by racing cross validation errors of competing subsets
        Specified by:
        search in class ASSearch
        Parameters:
        ASEval - the attribute evaluator to guide the search
        data - the training instances.
        Returns:
        an array (not necessarily ordered) of selected attribute indexes
        Throws:
        java.lang.Exception - if the search can't be completed
      • rankedAttributes

        public double[][] rankedAttributes()
                                    throws java.lang.Exception
        Description copied from interface: RankedOutputSearch
        Returns a X by 2 list of attribute indexes and corresponding evaluations from best (highest) to worst.
        Specified by:
        rankedAttributes in interface RankedOutputSearch
        Returns:
        the ranked list of attribute indexes in an array of ints
        Throws:
        java.lang.Exception - if the ranking can't be produced
      • toString

        public java.lang.String toString()
        Returns a string represenation
        Overrides:
        toString in class java.lang.Object
        Returns:
        a string representation