discretize.java

一个数据挖掘软件ALPHAMINERR的整个过程的JAVA版源代码

/*
 *    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.
 */

/*
 *    Discretize.java
 *    Copyright (C) 1999 Eibe Frank,Len Trigg
 *
 */


package weka.filters.unsupervised.attribute;

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

import weka.core.Attribute;
import weka.core.FastVector;
import weka.core.Instance;
import weka.core.Instances;
import weka.core.Option;
import weka.core.OptionHandler;
import weka.core.Range;
import weka.core.SparseInstance;
import weka.core.Utils;
import weka.core.WeightedInstancesHandler;
import weka.filters.Filter;
import weka.filters.UnsupervisedFilter;

/** 
 * An instance filter that discretizes a range of numeric attributes in 
 * the dataset into nominal attributes. Discretization is by simple binning.
 * Skips the class attribute if set.<p>
 *
 * Valid filter-specific options are: <p>
 *
 * -B num <br>
 * Specifies the (maximum) number of bins to divide numeric attributes into.
 * Default = 10.<p>
 *
 * -M num <br>
 * Specifies the desired weight of instances per bin for equal-frequency
 * binning. If this is set to a positive number then the -B option will be 
 * ignored. Default = -1.<p>
 *
 * -F <br>
 * Use equal-frequency instead of equal-width discretization if 
 * class-based discretisation is turned off.<p>
 *
 * -O <br>
 * Optimize the number of bins using a leave-one-out estimate of the 
 * entropy (for equal-width binning). If this is set then the -B option
 * will be ignored.<p>
 *
 * -R col1,col2-col4,... <br>
 * Specifies list of columns to Discretize. First
 * and last are valid indexes. (default: first-last) <p>
 *
 * -V <br>
 * Invert matching sense.<p>
 *
 * -D <br>
 * Make binary nominal attributes. <p>
 * 
 * @author Len Trigg ([e-mail])
 * @author Eibe Frank ([e-mail])
 * @version $Revision$
 */
public class Discretize extends PotentialClassIgnorer 
  implements UnsupervisedFilter, OptionHandler, WeightedInstancesHandler {

  /** Stores which columns to Discretize */
  protected Range m_DiscretizeCols = new Range();

  /** The number of bins to divide the attribute into */
  protected int m_NumBins = 10;

  /** The desired weight of instances per bin */
  protected double m_DesiredWeightOfInstancesPerInterval = -1;

  /** Store the current cutpoints */
  protected double [][] m_CutPoints = null;

  /** Output binary attributes for discretized attributes. */
  protected boolean m_MakeBinary = false;

  /** Find the number of bins using cross-validated entropy. */
  protected boolean m_FindNumBins = false;

  /** Use equal-frequency binning if unsupervised discretization turned on */
  protected boolean m_UseEqualFrequency = false;

  /** The default columns to discretize */
  protected String m_DefaultCols;

  /** Constructor - initialises the filter */
  public Discretize() {

    m_DefaultCols = "first-last";
    setAttributeIndices("first-last");
  }

  /** Another constructor */
  public Discretize(String cols) {

    m_DefaultCols = cols;
    setAttributeIndices(cols);
  }

  /**
   * Gets an enumeration describing the available options.
   *
   * @return an enumeration of all the available options.
   */
  public Enumeration listOptions() {

    Vector newVector = new Vector(7);

    newVector.addElement(new Option(
              "\tSpecifies the (maximum) number of bins to divide numeric"
	      + " attributes into.\n"
	      + "\t(default = 10)",
              "B", 1, "-B <num>"));

    newVector.addElement(new Option(
              "\tSpecifies the desired weight of instances per bin for\n"
	      + "\tequal-frequency binning. If this is set to a positive\n"
	      + "\tnumber then the -B option will be ignored.\n"
	      + "\t(default = -1)",
              "M", 1, "-M <num>"));

    newVector.addElement(new Option(
              "\tUse equal-frequency instead of equal-width discretization.",
              "F", 0, "-F"));

    newVector.addElement(new Option(
              "\tOptimize number of bins using leave-one-out estimate\n"+
	      "\tof estimated entropy (for equal-width discretization).\n"+
	      "\tIf this is set then the -B option will be ignored.",
              "O", 0, "-O"));

    newVector.addElement(new Option(
              "\tSpecifies list of columns to Discretize. First"
	      + " and last are valid indexes.\n"
	      + "\t(default: first-last)",
              "R", 1, "-R <col1,col2-col4,...>"));

    newVector.addElement(new Option(
              "\tInvert matching sense of column indexes.",
              "V", 0, "-V"));

    newVector.addElement(new Option(
              "\tOutput binary attributes for discretized attributes.",
              "D", 0, "-D"));

    return newVector.elements();
  }


  /**
   * Parses the options for this object. Valid options are: <p>
   *
   * -B num <br>
   * Specifies the (maximum) number of bins to divide numeric attributes into.
   * Default = 10.<p>
   *
   * -M num <br>
   * Specifies the desired weight of instances per bin for equal-frequency
   * binning. If this is set to a positive number then the -B option will be 
   * ignored. Default = -1.<p>
   *
   * -F <br>
   * Use equal-frequency instead of equal-width discretization if 
   * class-based discretisation is turned off.<p>
   *
   * -O <br>
   * Optimize the number of bins using a leave-one-out estimate of the 
   * entropy (for equal-width binning). If this is set then the -B
   * option will be ignored.<p>
   *
   * -R col1,col2-col4,... <br>
   * Specifies list of columns to Discretize. First
   * and last are valid indexes. (default none) <p>
   *
   * -V <br>
   * Invert matching sense.<p>
   *
   * -D <br>
   * Make binary nominal attributes. <p>
   * 
   * @param options the list of options as an array of strings
   * @exception Exception if an option is not supported
   */
  public void setOptions(String[] options) throws Exception {

    setMakeBinary(Utils.getFlag('D', options));
    setUseEqualFrequency(Utils.getFlag('F', options));
    setFindNumBins(Utils.getFlag('O', options));
    setInvertSelection(Utils.getFlag('V', options));

    String weight = Utils.getOption('M', options);
    if (weight.length() != 0) {
      setDesiredWeightOfInstancesPerInterval((new Double(weight)).doubleValue());
    } else {
      setDesiredWeightOfInstancesPerInterval(-1);
    }

    String numBins = Utils.getOption('B', options);
    if (numBins.length() != 0) {
      setBins(Integer.parseInt(numBins));
    } else {
      setBins(10);
    }
    
    String convertList = Utils.getOption('R', options);
    if (convertList.length() != 0) {
      setAttributeIndices(convertList);
    } else {
      setAttributeIndices(m_DefaultCols);
    }

    if (getInputFormat() != null) {
      setInputFormat(getInputFormat());
    }
  }

  /**
   * Gets the current settings of the filter.
   *
   * @return an array of strings suitable for passing to setOptions
   */
  public String [] getOptions() {

    String [] options = new String [10];
    int current = 0;

    if (getMakeBinary()) {
      options[current++] = "-D";
    }
    if (getUseEqualFrequency()) {
      options[current++] = "-F";
    }
    if (getFindNumBins()) {
      options[current++] = "-O";
    }
    if (getInvertSelection()) {
      options[current++] = "-V";
    }
    options[current++] = "-B"; options[current++] = "" + getBins();
    options[current++] = "-M"; 
    options[current++] = "" + getDesiredWeightOfInstancesPerInterval();
    if (!getAttributeIndices().equals("")) {
      options[current++] = "-R"; options[current++] = getAttributeIndices();
    }
    while (current < options.length) {
      options[current++] = "";
    }
    return options;
  }

  /**
   * Sets the format of the input instances.
   *
   * @param instanceInfo an Instances object containing the input instance
   * structure (any instances contained in the object are ignored - only the
   * structure is required).
   * @return true if the outputFormat may be collected immediately
   * @exception Exception if the input format can't be set successfully
   */
  public boolean setInputFormat(Instances instanceInfo) throws Exception {

    if (m_MakeBinary && m_IgnoreClass) {
      throw new IllegalArgumentException("Can't ignore class when " +
					 "changing the number of attributes!");
    }

    super.setInputFormat(instanceInfo);

    m_DiscretizeCols.setUpper(instanceInfo.numAttributes() - 1);
    m_CutPoints = null;
    
    if (getFindNumBins() && getUseEqualFrequency()) {
      throw new IllegalArgumentException("Bin number optimization in conjunction "+
					 "with equal-frequency binning not implemented.");
    }

    // If we implement loading cutfiles, then load 
    //them here and set the output format
    return false;
  }

  /**
   * Input an instance for filtering. Ordinarily the instance is processed
   * and made available for output immediately. Some filters require all
   * instances be read before producing output.
   *
   * @param instance the input instance
   * @return true if the filtered instance may now be
   * collected with output().
   * @exception IllegalStateException if no input format has been defined.
   */
  public boolean input(Instance instance) {

    if (getInputFormat() == null) {
      throw new IllegalStateException("No input instance format defined");
    }
    if (m_NewBatch) {
      resetQueue();
      m_NewBatch = false;
    }
    
    if (m_CutPoints != null) {
      convertInstance(instance);
      return true;
    }

    bufferInput(instance);
    return false;
  }

  /**
   * Signifies that this batch of input to the filter is finished. If the 
   * filter requires all instances prior to filtering, output() may now 
   * be called to retrieve the filtered instances.
   *
   * @return true if there are instances pending output
   * @exception IllegalStateException if no input structure has been defined
   */
  public boolean batchFinished() {

    if (getInputFormat() == null) {
      throw new IllegalStateException("No input instance format defined");
    }
    if (m_CutPoints == null) {
      calculateCutPoints();

      setOutputFormat();

      // If we implement saving cutfiles, save the cuts here

      // Convert pending input instances
      for(int i = 0; i < getInputFormat().numInstances(); i++) {
	convertInstance(getInputFormat().instance(i));
      }
    } 
    flushInput();

    m_NewBatch = true;
    return (numPendingOutput() != 0);
  }

  /**
   * Returns a string describing this filter
   *
   * @return a description of the filter suitable for
   * displaying in the explorer/experimenter gui
   */
  public String globalInfo() {

    return "An instance filter that discretizes a range of numeric"
      + " attributes in the dataset into nominal attributes."
      + " Discretization is by simple binning. Skips the class"
      + " attribute if set.";
  }
  
  /**
   * Returns the tip text for this property
   *
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String findNumBinsTipText() {

    return "Optimize number of equal-width bins using leave-one-out. Doesn't " +
      "work for equal-frequency binning";
  }

  /**
   * Get the value of FindNumBins.
   *
   * @return Value of FindNumBins.
   */
  public boolean getFindNumBins() {
    
    return m_FindNumBins;
  }
  
  /**
   * Set the value of FindNumBins.
   *
   * @param newFindNumBins Value to assign to FindNumBins.
   */
  public void setFindNumBins(boolean newFindNumBins) {
    
    m_FindNumBins = newFindNumBins;
  }
  
  /**
   * Returns the tip text for this property
   *
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String makeBinaryTipText() {

    return "Make resulting attributes binary.";
  }

  /**
   * Gets whether binary attributes should be made for discretized ones.
   *
   * @return true if attributes will be binarized
   */
  public boolean getMakeBinary() {

    return m_MakeBinary;
  }

  /** 
   * Sets whether binary attributes should be made for discretized ones.
   *
   * @param makeBinary if binary attributes are to be made
   */
  public void setMakeBinary(boolean makeBinary) {

    m_MakeBinary = makeBinary;
  }
  
  /**
   * Returns the tip text for this property
   *
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String desiredWeightOfInstancesPerIntervalTipText() {

    return "Sets the desired weight of instances per interval for " +
      "equal-frequency binning.";
  }
  
  /**
   * Get the DesiredWeightOfInstancesPerInterval value.
   * @return the DesiredWeightOfInstancesPerInterval value.
   */
  public double getDesiredWeightOfInstancesPerInterval() {

    return m_DesiredWeightOfInstancesPerInterval;
  }

  /**
   * Set the DesiredWeightOfInstancesPerInterval value.
   * @param newDesiredNumber The new DesiredNumber value.
   */
  public void setDesiredWeightOfInstancesPerInterval(double newDesiredNumber) {
    
    m_DesiredWeightOfInstancesPerInterval = newDesiredNumber;
  }
  
  /**
   * Returns the tip text for this property
   *
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String useEqualFrequencyTipText() {

    return "If set to true, equal-frequency binning will be used instead of" +
      " equal-width binning.";
  }
  
  /**
   * Get the value of UseEqualFrequency.
   *
   * @return Value of UseEqualFrequency.
   */
  public boolean getUseEqualFrequency() {
    
    return m_UseEqualFrequency;
  }
  
  /**
   * Set the value of UseEqualFrequency.
   *
   * @param newUseEqualFrequency Value to assign to UseEqualFrequency.
   */
  public void setUseEqualFrequency(boolean newUseEqualFrequency) {
    
    m_UseEqualFrequency = newUseEqualFrequency;
  }

  /**
   * Returns the tip text for this property
   *
   * @return tip text for this property suitable for
   * displaying in the explorer/experimenter gui
   */
  public String binsTipText() {

    return "Number of bins.";
  }

  /**
   * Gets the number of bins numeric attributes will be divided into