aprioriitemset.java

MacroWeka扩展了著名数据挖掘工具weka

/*
 *    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.
 */

/*
 *    AprioriItemSet.java
 *    Copyright (C) 2004 Eibe Frank, Stefan Mutter
 *
 */

package weka.associations;

import java.io.*;
import java.util.*;
import weka.core.*;


/**
 * Class for storing a set of items. Item sets are stored in a lexicographic
 * order, which is determined by the header information of the set of instances
 * used for generating the set of items. All methods in this class assume that
 * item sets are stored in lexicographic order.
 * The class provides methods that are used in the Apriori algorithm to construct
 * association rules.
 *
 * @author Eibe Frank ([e-mail]), Stefan Mutter ([e-mail])
 * @version $Revision: 1.1 $
 */
public class AprioriItemSet extends ItemSet implements Serializable {

  /**
   * Constructor
   * @param totalTrans the total number of transactions in the data
   */
  public AprioriItemSet(int totalTrans) {
    super(totalTrans);
  }

  
  /**
   * Outputs the confidence for a rule.
   *
   * @param premise the premise of the rule
   * @param consequence the consequence of the rule
   * @return the confidence on the training data
   */
  public static double confidenceForRule(AprioriItemSet premise, 
					 AprioriItemSet consequence) {

    return (double)consequence.m_counter/(double)premise.m_counter;
  }

  /**
   * Outputs the lift for a rule. Lift is defined as:<br>
   * confidence / prob(consequence)
   *
   * @param premise the premise of the rule
   * @param consequence the consequence of the rule
   * @param consequenceCount how many times the consequence occurs independent
   * of the premise
   * @return the lift on the training data
   */
  public double liftForRule(AprioriItemSet premise, 
			    AprioriItemSet consequence,
			    int consequenceCount) {
    double confidence = confidenceForRule(premise, consequence);

   return confidence / ((double)consequenceCount / 
	  (double)m_totalTransactions);
  }

  /**
   * Outputs the leverage for a rule. Leverage is defined as: <br>
   * prob(premise & consequence) - (prob(premise) * prob(consequence))
   *
   * @param premise the premise of the rule
   * @param consequence the consequence of the rule
   * @param premiseCount how many times the premise occurs independent
   * of the consequent
   * @param consequenceCount how many times the consequence occurs independent
   * of the premise
   * @return the leverage on the training data
   */
  public double leverageForRule(AprioriItemSet premise,
				AprioriItemSet consequence,
				int premiseCount,
				int consequenceCount) {
    double coverageForItemSet = (double)consequence.m_counter / 
      (double)m_totalTransactions;
    double expectedCoverageIfIndependent = 
      ((double)premiseCount / (double)m_totalTransactions) * 
      ((double)consequenceCount / (double)m_totalTransactions);
    double lev = coverageForItemSet - expectedCoverageIfIndependent;
    return lev;
  }

  /**
   * Outputs the conviction for a rule. Conviction is defined as: <br>
   * prob(premise) * prob(!consequence) / prob(premise & !consequence)
   *
   * @param premise the premise of the rule
   * @param consequence the consequence of the rule
   * @param premiseCount how many times the premise occurs independent
   * of the consequent
   * @param consequenceCount how many times the consequence occurs independent
   * of the premise
   * @return the conviction on the training data
   */
  public double convictionForRule(AprioriItemSet premise,
				   AprioriItemSet consequence,
				   int premiseCount,
				   int consequenceCount) {
    double num = 
      (double)premiseCount * (double)(m_totalTransactions - consequenceCount) /
       (double)m_totalTransactions;
    double denom = 
      ((premiseCount - consequence.m_counter)+1);
    
    if (num < 0 || denom < 0) {
      System.err.println("*** "+num+" "+denom);
      System.err.println("premis count: "+premiseCount+" consequence count "+consequenceCount+" total trans "+m_totalTransactions);
    }
    return num / denom;
  }

  
  
  /**
   * Generates all rules for an item set.
   *
   * @param minConfidence the minimum confidence the rules have to have
   * @param hashtables containing all(!) previously generated
   * item sets
   * @param numItemsInSet the size of the item set for which the rules
   * are to be generated
   * @return all the rules with minimum confidence for the given item set
   */
  public FastVector[] generateRules(double minConfidence, 
					  FastVector hashtables,
					  int numItemsInSet) {

    FastVector premises = new FastVector(),consequences = new FastVector(),
      conf = new FastVector();
    FastVector[] rules = new FastVector[3], moreResults;
    AprioriItemSet premise, consequence;
    Hashtable hashtable = (Hashtable)hashtables.elementAt(numItemsInSet - 2);

    // Generate all rules with one item in the consequence.
    for (int i = 0; i < m_items.length; i++) 
      if (m_items[i] != -1) {
	premise = new AprioriItemSet(m_totalTransactions);
	consequence = new AprioriItemSet(m_totalTransactions);
	premise.m_items = new int[m_items.length];
	consequence.m_items = new int[m_items.length];
	consequence.m_counter = m_counter;
        
	for (int j = 0; j < m_items.length; j++) 
	  consequence.m_items[j] = -1;
	System.arraycopy(m_items, 0, premise.m_items, 0, m_items.length);
	premise.m_items[i] = -1;
        
	consequence.m_items[i] = m_items[i];
	premise.m_counter = ((Integer)hashtable.get(premise)).intValue();
	premises.addElement(premise);
	consequences.addElement(consequence);
	conf.addElement(new Double(confidenceForRule(premise, consequence)));
      }
    rules[0] = premises;
    rules[1] = consequences;
    rules[2] = conf;
    pruneRules(rules, minConfidence);

    // Generate all the other rules
    moreResults = moreComplexRules(rules, numItemsInSet, 1, minConfidence,
				   hashtables);
    if (moreResults != null) 
      for (int i = 0; i < moreResults[0].size(); i++) {
	rules[0].addElement(moreResults[0].elementAt(i));
	rules[1].addElement(moreResults[1].elementAt(i));
	rules[2].addElement(moreResults[2].elementAt(i));
      }
    return rules;
  }


  
  /**
   * Generates all significant rules for an item set.
   *
   * @param minMetric the minimum metric (confidence, lift, leverage, 
   * improvement) the rules have to have
   * @param metricType (confidence=0, lift, leverage, improvement)
   * @param hashtables containing all(!) previously generated
   * item sets
   * @param numItemsInSet the size of the item set for which the rules
   * are to be generated
   * @param the significance level for testing the rules
   * @return all the rules with minimum metric for the given item set
   * @exception Exception if something goes wrong
   */
  public final FastVector[] generateRulesBruteForce(double minMetric,
						    int metricType,
						FastVector hashtables,
						int numItemsInSet,
						int numTransactions,
						double significanceLevel) 
  throws Exception {

    FastVector premises = new FastVector(),consequences = new FastVector(),
      conf = new FastVector(), lift = new FastVector(), lev = new FastVector(),
      conv = new FastVector(); 
    FastVector[] rules = new FastVector[6];
    AprioriItemSet premise, consequence;
    Hashtable hashtableForPremise, hashtableForConsequence;
    int numItemsInPremise, help, max, consequenceUnconditionedCounter;
    double[][] contingencyTable = new double[2][2];
    double metric, chiSquared;

    // Generate all possible rules for this item set and test their
    // significance.
    max = (int)Math.pow(2, numItemsInSet);
    for (int j = 1; j < max; j++) {
      numItemsInPremise = 0;
      help = j;
      while (help > 0) {
	if (help % 2 == 1)
	  numItemsInPremise++;
	help /= 2;
      }
      if (numItemsInPremise < numItemsInSet) {
	hashtableForPremise = 
	  (Hashtable)hashtables.elementAt(numItemsInPremise-1);
	hashtableForConsequence = 
	  (Hashtable)hashtables.elementAt(numItemsInSet-numItemsInPremise-1);
	premise = new AprioriItemSet(m_totalTransactions);
	consequence = new AprioriItemSet(m_totalTransactions);
	premise.m_items = new int[m_items.length];
        
	consequence.m_items = new int[m_items.length];
	consequence.m_counter = m_counter;
	help = j;
	for (int i = 0; i < m_items.length; i++) 
	  if (m_items[i] != -1) {
	    if (help % 2 == 1) {          
	      premise.m_items[i] = m_items[i];
	      consequence.m_items[i] = -1;
	    } else {
	      premise.m_items[i] = -1;
	      consequence.m_items[i] = m_items[i];
	    }
	    help /= 2;
	  } else {