subspacecluster.java

代码是一个分类器的实现,其中使用了部分weka的源代码。可以将项目导入eclipse运行

/*
 *    This program is free software; you can redistribute it and/or modify
 *    it under the terms of the GNU General Public License as published by
 *    the Free Software Foundation; either version 2 of the License, or
 *    (at your option) any later version.
 *
 *    This program is distributed in the hope that it will be useful,
 *    but WITHOUT ANY WARRANTY; without even the implied warranty of
 *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *    GNU General Public License for more details.
 *
 *    You should have received a copy of the GNU General Public License
 *    along with this program; if not, write to the Free Software
 *    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

/*
 *    SubspaceCluster.java
 *    Copyright (C) 2001 Gabi Schmidberger
 *
 */

package weka.datagenerators.clusterers;

import weka.core.Attribute;
import weka.core.FastVector;
import weka.core.Instance;
import weka.core.Instances;
import weka.core.Option;
import weka.core.Range;
import weka.core.Tag;
import weka.core.Utils;
import weka.datagenerators.ClusterDefinition;
import weka.datagenerators.ClusterGenerator;

import java.util.Enumeration;
import java.util.Random;
import java.util.Vector;

/**
 <!-- globalinfo-start -->
 * A data generator that produces data points in hyperrectangular subspace clusters.
 * <p/>
 <!-- globalinfo-end -->
 *
 <!-- options-start -->
 * Valid options are: <p/>
 * 
 * <pre> -h
 *  Prints this help.</pre>
 * 
 * <pre> -o &lt;file&gt;
 *  The name of the output file, otherwise the generated data is
 *  printed to stdout.</pre>
 * 
 * <pre> -r &lt;name&gt;
 *  The name of the relation.</pre>
 * 
 * <pre> -d
 *  Whether to print debug informations.</pre>
 * 
 * <pre> -S
 *  The seed for random function (default 1)</pre>
 * 
 * <pre> -a &lt;num&gt;
 *  The number of attributes (default 1).</pre>
 * 
 * <pre> -c
 *  Class Flag, if set, the cluster is listed in extra attribute.</pre>
 * 
 * <pre> -b &lt;range&gt;
 *  The indices for boolean attributes.</pre>
 * 
 * <pre> -m &lt;range&gt;
 *  The indices for nominal attributes.</pre>
 * 
 * <pre> -P &lt;num&gt;
 *  The noise rate in percent (default 0.0).
 *  Can be between 0% and 30%. (Remark: The original 
 *  algorithm only allows noise up to 10%.)</pre>
 * 
 * <pre> -C &lt;cluster-definition&gt;
 *  A cluster definition of class 'SubspaceClusterDefinition'
 *  (definition needs to be quoted to be recognized as 
 *  a single argument).</pre>
 * 
 * <pre> 
 * Options specific to weka.datagenerators.clusterers.SubspaceClusterDefinition:
 * </pre>
 * 
 * <pre> -A &lt;range&gt;
 *  Generates randomly distributed instances in the cluster.</pre>
 * 
 * <pre> -U &lt;range&gt;
 *  Generates uniformly distributed instances in the cluster.</pre>
 * 
 * <pre> -G &lt;range&gt;
 *  Generates gaussian distributed instances in the cluster.</pre>
 * 
 * <pre> -D &lt;num&gt;,&lt;num&gt;
 *  The attribute min/max (-A and -U) or mean/stddev (-G) for
 *  the cluster.</pre>
 * 
 * <pre> -N &lt;num&gt;..&lt;num&gt;
 *  The range of number of instances per cluster (default 1..50).</pre>
 * 
 * <pre> -I
 *  Uses integer instead of continuous values (default continuous).</pre>
 * 
 <!-- options-end -->
 *
 * @author Gabi Schmidberger ([e-mail])
 * @author  FracPete (fracpete at waikato dot ac dot nz)
 * @version $Revision: 1.3 $ 
 */
public class SubspaceCluster 
  extends ClusterGenerator {

  /** for serialization */
  static final long serialVersionUID = -3454999858505621128L;
  
  /** noise rate in percent (option P,  between 0 and 30)*/ 
  protected double m_NoiseRate;

  /** cluster list */
  protected ClusterDefinition[] m_Clusters;

  /** if nominal, store number of values */
  protected int[] m_numValues;

  /** store global min values */
  protected double[] m_globalMinValue;

  /** store global max values */
  protected double[] m_globalMaxValue;

  /** cluster type: uniform/random */
  public static final int UNIFORM_RANDOM = 0;  
  /** cluster type: total uniform */
  public static final int TOTAL_UNIFORM = 1;
  /** cluster type: gaussian */
  public static final int GAUSSIAN = 2;
  /** the tags for the cluster types */
  public static final Tag[] TAGS_CLUSTERTYPE = {
    new Tag(UNIFORM_RANDOM, "uniform/random"),
    new Tag(TOTAL_UNIFORM,  "total uniform"),
    new Tag(GAUSSIAN,       "gaussian")
  };

  /** cluster subtype: continuous */
  public static final int CONTINUOUS = 0;
  /** cluster subtype: integer */
  public static final int INTEGER = 1;
  /** the tags for the cluster types */
  public static final Tag[] TAGS_CLUSTERSUBTYPE = {
    new Tag(CONTINUOUS, "continuous"),
    new Tag(INTEGER,    "integer")
  };

  /**
   * initializes the generator, sets the number of clusters to 0, since user
   * has to specify them explicitly
   */
  public SubspaceCluster() {
    super();

    setNoiseRate(defaultNoiseRate());
  }

  /**
   * Returns a string describing this data generator.
   *
   * @return a description of the data generator suitable for
   * displaying in the explorer/experimenter gui
   */
  public String globalInfo() {
    return "A data generator that produces data points in "
      + "hyperrectangular subspace clusters.";
  }

  /**
   * Returns an enumeration describing the available options.
   *
   * @return an enumeration of all the available options
   */
  public Enumeration listOptions() {
    Vector result = enumToVector(super.listOptions());

    result.addElement(new Option(
          "\tThe noise rate in percent (default " 
          + defaultNoiseRate() + ").\n"
          + "\tCan be between 0% and 30%. (Remark: The original \n"
          + "\talgorithm only allows noise up to 10%.)",
          "P", 1, "-P <num>"));

    result.addElement(new Option(
          "\tA cluster definition of class '" 
	  + SubspaceClusterDefinition.class.getName().replaceAll(".*\\.", "") + "'\n"
	  + "\t(definition needs to be quoted to be recognized as \n"
	  + "\ta single argument).",
          "C", 1, "-C <cluster-definition>"));

    result.addElement(new Option(
	      "", "", 0, 
	      "\nOptions specific to " 
	      + SubspaceClusterDefinition.class.getName() + ":"));

    result.addAll(
        enumToVector(new SubspaceClusterDefinition(this).listOptions()));

    return result.elements();
  }

  /**
   * Parses a list of options for this object. <p/>
   *
   <!-- options-start -->
   * Valid options are: <p/>
   * 
   * <pre> -h
   *  Prints this help.</pre>
   * 
   * <pre> -o &lt;file&gt;
   *  The name of the output file, otherwise the generated data is
   *  printed to stdout.</pre>
   * 
   * <pre> -r &lt;name&gt;
   *  The name of the relation.</pre>
   * 
   * <pre> -d
   *  Whether to print debug informations.</pre>
   * 
   * <pre> -S
   *  The seed for random function (default 1)</pre>
   * 
   * <pre> -a &lt;num&gt;
   *  The number of attributes (default 1).</pre>
   * 
   * <pre> -c
   *  Class Flag, if set, the cluster is listed in extra attribute.</pre>
   * 
   * <pre> -b &lt;range&gt;
   *  The indices for boolean attributes.</pre>
   * 
   * <pre> -m &lt;range&gt;
   *  The indices for nominal attributes.</pre>
   * 
   * <pre> -P &lt;num&gt;
   *  The noise rate in percent (default 0.0).
   *  Can be between 0% and 30%. (Remark: The original 
   *  algorithm only allows noise up to 10%.)</pre>
   * 
   * <pre> -C &lt;cluster-definition&gt;
   *  A cluster definition of class 'SubspaceClusterDefinition'
   *  (definition needs to be quoted to be recognized as 
   *  a single argument).</pre>
   * 
   * <pre> 
   * Options specific to weka.datagenerators.clusterers.SubspaceClusterDefinition:
   * </pre>
   * 
   * <pre> -A &lt;range&gt;
   *  Generates randomly distributed instances in the cluster.</pre>
   * 
   * <pre> -U &lt;range&gt;
   *  Generates uniformly distributed instances in the cluster.</pre>
   * 
   * <pre> -G &lt;range&gt;
   *  Generates gaussian distributed instances in the cluster.</pre>
   * 
   * <pre> -D &lt;num&gt;,&lt;num&gt;
   *  The attribute min/max (-A and -U) or mean/stddev (-G) for
   *  the cluster.</pre>
   * 
   * <pre> -N &lt;num&gt;..&lt;num&gt;
   *  The range of number of instances per cluster (default 1..50).</pre>
   * 
   * <pre> -I
   *  Uses integer instead of continuous values (default continuous).</pre>
   * 
   <!-- options-end -->
   *
   * @param options the list of options as an array of strings
   * @throws Exception if an option is not supported
   */
  public void setOptions(String[] options) throws Exception {
    String                      tmpStr;
    SubspaceClusterDefinition   cl;
    Vector                      list;
    int                         clCount;

    super.setOptions(options);

    m_numValues = new int[getNumAttributes()];
    // numValues might be changed by a cluster definition
    // (only relevant for nominal data)
    for (int i = 0; i < getNumAttributes(); i++)
      m_numValues[i] = 1;

    tmpStr = Utils.getOption('P', options);
    if (tmpStr.length() != 0)
      setNoiseRate(Double.parseDouble(tmpStr));
    else
      setNoiseRate(defaultNoiseRate());

    // cluster definitions
    list = new Vector();
    
    clCount = 0;
    do {
      tmpStr = Utils.getOption('C', options);
      if (tmpStr.length() != 0) {
        clCount++;
        cl = new SubspaceClusterDefinition(this);
        cl.setOptions(Utils.splitOptions(tmpStr));
        list.add(cl);
      }
    }
    while (tmpStr.length() != 0);

    m_Clusters = (ClusterDefinition[]) 
                    list.toArray(new ClusterDefinition[list.size()]);
    // in case no cluster definition was provided, make sure that there's at
    // least one definition present -> see getClusters()
    getClusters();
  }


  /**
   * Gets the current settings of the datagenerator.
   *
   * @return an array of strings suitable for passing to setOptions
   */
  public String[] getOptions() {
    Vector        result;
    String[]      options;
    int           i;

    result  = new Vector();
    options = super.getOptions();
    for (i = 0; i < options.length; i++)
      result.add(options[i]);

    result.add("-P"); 
    result.add("" + getNoiseRate());

    for (i = 0; i < getClusters().length; i++)  {
      result.add("-C");
      result.add(Utils.joinOptions(getClusters()[i].getOptions()));
    }

    return (String[]) result.toArray(new String[result.size()]);
  }

  /**
   * returns the current cluster definitions, if necessary initializes them
   * 
   * @return the current cluster definitions
   */
  protected ClusterDefinition[] getClusters() {
    if ( (m_Clusters == null) || (m_Clusters.length == 0) ) {
      if (m_Clusters != null)
        System.out.println("NOTE: at least 1 cluster definition is necessary, " 
            + "created default one.");
      m_Clusters = new ClusterDefinition[]{new SubspaceClusterDefinition(this)};
    }

    return m_Clusters;
  }

  /**
   * returns the default number of attributes
   * 
   * @return the default number of attributes
   */
  protected int defaultNumAttributes() {
    return 1;
  }

  /**
   * Sets the number of attributes the dataset should have.
   * @param numAttributes the new number of attributes
   */
  public void setNumAttributes(int numAttributes) {
    super.setNumAttributes(numAttributes);
    m_numValues = new int[getNumAttributes()];
  }
  
  /**
   * Returns the tip text for this property
   * 
   * @return tip text for this property suitable for
   *         displaying in the explorer/experimenter gui
   */
  public String numAttributesTipText() {
    return "The number of attributes the generated data will contain (Note: they must be covered by the cluster definitions!)";
  }

  /**
   * returns the default noise rate
   * 
   * @return the default noise rate
   */
  protected double defaultNoiseRate() {
    return 0.0;
  }

  /**
   * Gets the percentage of noise set.
   *
   * @return the percentage of noise set
   */
  public double getNoiseRate() { 
    return m_NoiseRate; 
  }

  /**
   * Sets the percentage of noise set.
   *
   * @param newNoiseRate new percentage of noise 
   */
  public void setNoiseRate(double newNoiseRate) {
    m_NoiseRate = newNoiseRate;
  }

  /**
   * Returns the tip text for this property
   * 
   * @return tip text for this property suitable for
   *         displaying in the explorer/experimenter gui
   */
  public String noiseRateTipText() {
    return "The noise rate to use.";
  }

  /**
   * returns the currently set clusters
   * 
   * @return the currently set clusters
   */
  public ClusterDefinition[] getClusterDefinitions() {
    return getClusters();
  }

  /**
   * sets the clusters to use
   * 
   * @param value the clusters do use
   * @throws Exception if clusters are not the correct class
   */
  public void setClusterDefinitions(ClusterDefinition[] value) 
    throws Exception {

    String      indexStr;
    
    indexStr   = "";
    m_Clusters = value;
    for (int i = 0; i < getClusters().length; i++) {
      if (!(getClusters()[i] instanceof SubspaceClusterDefinition)) {
        if (indexStr.length() != 0)
          indexStr += ",";
        indexStr += "" + (i+1);
      }
      getClusters()[i].setParent(this);
      getClusters()[i].setOptions(getClusters()[i].getOptions()); // for initializing!
    }

    // any wrong classes encountered?
    if (indexStr.length() != 0)
      throw new Exception("These cluster definitions are not '" 
          + SubspaceClusterDefinition.class.getName() + "': " + indexStr);
  }

  /**
   * Returns the tip text for this property
   * 
   * @return tip text for this property suitable for
   *         displaying in the explorer/experimenter gui
   */
  public String clusterDefinitionsTipText() {
    return "The clusters to use.";
  }

  /**
   * Checks, whether all attributes are covered by cluster definitions and 
   * returns TRUE in that case.
   * 
   * @return whether all attributes are covered
   */
  protected boolean checkCoverage() {
    int         i;
    int         n;
    int[]       count;
    Range       r;
    String      attrIndex;
    SubspaceClusterDefinition  cl;