learningrateresultproducer.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.
 */

/*
 *    LearningRateResultProducer.java
 *    Copyright (C) 1999 Len Trigg
 *
 */


package weka.experiment;

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

import weka.core.AdditionalMeasureProducer;
import weka.core.Instances;
import weka.core.Option;
import weka.core.OptionHandler;
import weka.core.Utils;

/**
 * LearningRateResultProducer takes the results from a ResultProducer
 * and submits the average to the result listener. For non-numeric
 * result fields, the first value is used.
 *
 * @author Len Trigg ([e-mail])
 * @version $Revision$
 */
public class LearningRateResultProducer 
  implements ResultListener, ResultProducer, OptionHandler,
	     AdditionalMeasureProducer {

  /** The dataset of interest */
  protected Instances m_Instances;

  /** The ResultListener to send results to */
  protected ResultListener m_ResultListener = new CSVResultListener();

  /** The ResultProducer used to generate results */
  protected ResultProducer m_ResultProducer
    = new AveragingResultProducer();

  /** The names of any additional measures to look for in SplitEvaluators */
  protected String [] m_AdditionalMeasures = null;

  /** 
   * The minimum number of instances to use. If this is zero, the first
   * step will contain m_StepSize instances 
   */
  protected int m_LowerSize = 0;
  
  /**
   * The maximum number of instances to use. -1 indicates no maximum 
   * (other than the total number of instances)
   */
  protected int m_UpperSize = -1;

  /** The number of instances to add at each step */
  protected int m_StepSize = 10;

  /** The current dataset size during stepping */
  protected int m_CurrentSize = 0;

  /* The name of the key field containing the learning rate step number */
  public static String STEP_FIELD_NAME = "Total_instances";

  /**
   * Returns a string describing this result producer
   * @return a description of the result producer suitable for
   * displaying in the explorer/experimenter gui
   */
  public String globalInfo() {
    return "Tells a sub-ResultProducer to reproduce the current run for "
      +"varying sized subsamples of the dataset. Normally used with "
      +"an AveragingResultProducer and CrossValidationResultProducer "
      +"combo to generate learning curve results.";
  }


  /**
   * Determines if there are any constraints (imposed by the
   * destination) on the result columns to be produced by
   * resultProducers. Null should be returned if there are NO
   * constraints, otherwise a list of column names should be
   * returned as an array of Strings.
   * @param rp the ResultProducer to which the constraints will apply
   * @return an array of column names to which resutltProducer's
   * results will be restricted.
   * @exception Exception if constraints can't be determined
   */
  public String [] determineColumnConstraints(ResultProducer rp) 
    throws Exception {
    return null;
  }

  /**
   * Gets the keys for a specified run number. Different run
   * numbers correspond to different randomizations of the data. Keys
   * produced should be sent to the current ResultListener
   *
   * @param run the run number to get keys for.
   * @exception Exception if a problem occurs while getting the keys
   */
  public void doRunKeys(int run) throws Exception {

    if (m_ResultProducer == null) {
      throw new Exception("No ResultProducer set");
    }
    if (m_ResultListener == null) {
      throw new Exception("No ResultListener set");
    }
    if (m_Instances == null) {
      throw new Exception("No Instances set");
    }

    // Tell the resultproducer to send results to us
    m_ResultProducer.setResultListener(this);
    m_ResultProducer.setInstances(m_Instances);

    // For each subsample size
    if (m_LowerSize == 0) {
      m_CurrentSize = m_StepSize;
    } else {
      m_CurrentSize = m_LowerSize;
    }
    while (m_CurrentSize <= m_Instances.numInstances() &&
           ((m_UpperSize == -1) ||
            (m_CurrentSize <= m_UpperSize))) {
      m_ResultProducer.doRunKeys(run);
      m_CurrentSize += m_StepSize;
    }
  }


  /**
   * Gets the results for a specified run number. Different run
   * numbers correspond to different randomizations of the data. Results
   * produced should be sent to the current ResultListener
   *
   * @param run the run number to get results for.
   * @exception Exception if a problem occurs while getting the results
   */
  public void doRun(int run) throws Exception {

    if (m_ResultProducer == null) {
      throw new Exception("No ResultProducer set");
    }
    if (m_ResultListener == null) {
      throw new Exception("No ResultListener set");
    }
    if (m_Instances == null) {
      throw new Exception("No Instances set");
    }

    // Randomize on a copy of the original dataset
    Instances runInstances = new Instances(m_Instances);
    runInstances.randomize(new Random(run));
    if (runInstances.classAttribute().isNominal()) {
      runInstances.stratify(m_StepSize);
    }

    // Tell the resultproducer to send results to us
    m_ResultProducer.setResultListener(this);

    // For each subsample size
    if (m_LowerSize == 0) {
      m_CurrentSize = m_StepSize;
    } else {
      m_CurrentSize = m_LowerSize;
    }
    while (m_CurrentSize <= m_Instances.numInstances() &&
           ((m_UpperSize == -1) ||
            (m_CurrentSize <= m_UpperSize))) {
      m_ResultProducer.setInstances(new Instances(runInstances, 0, 
                                                  m_CurrentSize));
      m_ResultProducer.doRun(run);
      m_CurrentSize += m_StepSize;
    }
  }

  
  
  /**
   * Prepare for the results to be received.
   *
   * @param rp the ResultProducer that will generate the results
   * @exception Exception if an error occurs during preprocessing.
   */
  public void preProcess(ResultProducer rp) throws Exception {

    if (m_ResultListener == null) {
      throw new Exception("No ResultListener set");
    }
    m_ResultListener.preProcess(this);
  }

  /**
   * Prepare to generate results. The ResultProducer should call
   * preProcess(this) on the ResultListener it is to send results to.
   *
   * @exception Exception if an error occurs during preprocessing.
   */
  public void preProcess() throws Exception {
    
    if (m_ResultProducer == null) {
      throw new Exception("No ResultProducer set");
    }
    // Tell the resultproducer to send results to us
    m_ResultProducer.setResultListener(this);
    m_ResultProducer.preProcess();
  }
  
  /**
   * When this method is called, it indicates that no more results
   * will be sent that need to be grouped together in any way.
   *
   * @param rp the ResultProducer that generated the results
   * @exception Exception if an error occurs
   */
  public void postProcess(ResultProducer rp) throws Exception {

    m_ResultListener.postProcess(this);
  }

  /**
   * When this method is called, it indicates that no more requests to
   * generate results for the current experiment will be sent. The
   * ResultProducer should call preProcess(this) on the
   * ResultListener it is to send results to.
   *
   * @exception Exception if an error occurs
   */
  public void postProcess() throws Exception {

    m_ResultProducer.postProcess();
  }
  
  /**
   * Accepts results from a ResultProducer.
   *
   * @param rp the ResultProducer that generated the results
   * @param key an array of Objects (Strings or Doubles) that uniquely
   * identify a result for a given ResultProducer with given compatibilityState
   * @param result the results stored in an array. The objects stored in
   * the array may be Strings, Doubles, or null (for the missing value).
   * @exception Exception if the result could not be accepted.
   */
  public void acceptResult(ResultProducer rp, Object [] key, Object [] result)
    throws Exception {

    if (m_ResultProducer != rp) {
      throw new Error("Unrecognized ResultProducer sending results!!");
    }
    // Add in current step as key field
    Object [] newKey = new Object [key.length + 1];
    System.arraycopy(key, 0, newKey, 0, key.length);
    newKey[key.length] = new String("" + m_CurrentSize);
    // Pass on to result listener
    m_ResultListener.acceptResult(this, newKey, result);
  }

  /**
   * Determines whether the results for a specified key must be
   * generated.
   *
   * @param rp the ResultProducer wanting to generate the results
   * @param key an array of Objects (Strings or Doubles) that uniquely
   * identify a result for a given ResultProducer with given compatibilityState
   * @return true if the result should be generated
   * @exception Exception if it could not be determined if the result 
   * is needed.
   */
  public boolean isResultRequired(ResultProducer rp, Object [] key) 
    throws Exception {

    if (m_ResultProducer != rp) {
      throw new Error("Unrecognized ResultProducer sending results!!");
    }
    // Add in current step as key field
    Object [] newKey = new Object [key.length + 1];
    System.arraycopy(key, 0, newKey, 0, key.length);
    newKey[key.length] = new String("" + m_CurrentSize);
    // Pass on request to result listener
    return m_ResultListener.isResultRequired(this, newKey);
  }

  /**
   * Gets the names of each of the columns produced for a single run.
   *
   * @return an array containing the name of each column
   * @exception Exception if key names cannot be generated
   */
  public String [] getKeyNames() throws Exception {

    String [] keyNames = m_ResultProducer.getKeyNames();
    String [] newKeyNames = new String [keyNames.length + 1];
    System.arraycopy(keyNames, 0, newKeyNames, 0, keyNames.length);
    // Think of a better name for this key field
    newKeyNames[keyNames.length] = STEP_FIELD_NAME;
    return newKeyNames;
  }

  /**
   * Gets the data types of each of the columns produced for a single run.
   * This method should really be static.
   *
   * @return an array containing objects of the type of each column. The 
   * objects should be Strings, or Doubles.
   * @exception Exception if the key types could not be determined (perhaps
   * because of a problem from a nested sub-resultproducer)
   */
  public Object [] getKeyTypes() throws Exception {

    Object [] keyTypes = m_ResultProducer.getKeyTypes();
    Object [] newKeyTypes = new Object [keyTypes.length + 1];
    System.arraycopy(keyTypes, 0, newKeyTypes, 0, keyTypes.length);
    newKeyTypes[keyTypes.length] = "";
    return newKeyTypes;
  }

  /**
   * Gets the names of each of the columns produced for a single run.
   * A new result field is added for the number of results used to
   * produce each average.
   * If only averages are being produced the names are not altered, if
   * standard deviations are produced then "Dev_" and "Avg_" are prepended
   * to each result deviation and average field respectively.
   *
   * @return an array containing the name of each column
   * @exception Exception if the result names could not be determined (perhaps
   * because of a problem from a nested sub-resultproducer)
   */
  public String [] getResultNames() throws Exception {

    return m_ResultProducer.getResultNames();
  }

  /**
   * Gets the data types of each of the columns produced for a single run.
   *
   * @return an array containing objects of the type of each column. The 
   * objects should be Strings, or Doubles.
   * @exception Exception if the result types could not be determined (perhaps
   * because of a problem from a nested sub-resultproducer)
   */
  public Object [] getResultTypes() throws Exception {

    return m_ResultProducer.getResultTypes();
  }

  /**
   * Gets a description of the internal settings of the result
   * producer, sufficient for distinguishing a ResultProducer
   * instance from another with different settings (ignoring
   * those settings set through this interface). For example,
   * a cross-validation ResultProducer may have a setting for the
   * number of folds. For a given state, the results produced should
   * be compatible. Typically if a ResultProducer is an OptionHandler,