sequentialbuild.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.
 */

 /**
  * Title: XELOPES Data Mining Library
  * Description: The XELOPES library is an open platform-independent and data-source-independent library for Embedded Data Mining.
  * Copyright: Copyright (c) 2002 Prudential Systems Software GmbH
  * Company: ZSoft (www.zsoft.ru), Prudsys (www.prudsys.com)
  * @author Carsten Weisse
  * @author Michael Thess
  * @version 1.0
  */

package com.prudsys.pdm.Examples;

import java.io.FileWriter;
import java.util.Vector;

import com.prudsys.pdm.Automat.MiningAutomationAssignment;
import com.prudsys.pdm.Core.CategoricalAttribute;
import com.prudsys.pdm.Core.Category;
import com.prudsys.pdm.Core.MiningAlgorithmSpecification;
import com.prudsys.pdm.Core.MiningDataSpecification;
import com.prudsys.pdm.Core.MiningException;
import com.prudsys.pdm.Core.MiningModel;
import com.prudsys.pdm.Core.NumericAttribute;
import com.prudsys.pdm.Input.MiningInputStream;
import com.prudsys.pdm.Input.Records.Csv.MiningCsvStream;
import com.prudsys.pdm.Models.Sequential.ItemSetSeq;
import com.prudsys.pdm.Models.Sequential.RuleSetSeq;
import com.prudsys.pdm.Models.Sequential.SequenceNumberAssessment;
import com.prudsys.pdm.Models.Sequential.SequenceNumberCallback;
import com.prudsys.pdm.Models.Sequential.SequentialAlgorithm;
import com.prudsys.pdm.Models.Sequential.SequentialMiningModel;
import com.prudsys.pdm.Models.Sequential.SequentialSettings;
import com.prudsys.pdm.Utils.GeneralUtils;
import com.prudsys.pdm.Utils.PmmlUtils;

/**
 * Builds a sequence analysis model and writes it to
 * PMML file 'SequenceModel.xml'.
 */
public class SequentialBuild extends BasisExample {

  /**
   * Empty constructor.
   */
  public SequentialBuild() {
  }

  /**
   * Run the example of this class.
   *
   * @throws Exception error while example is running
   */
  public void runExample() throws Exception {

    // Create metadata and open csv source stream:
    MiningDataSpecification mds = new MiningDataSpecification();
    mds.setRelationName("Sessions");
    CategoricalAttribute transactId = new CategoricalAttribute("transactionId");
    CategoricalAttribute itemId = new CategoricalAttribute("itemId");
    NumericAttribute itemIndex = new NumericAttribute("itemIndex");
    mds.addMiningAttribute(transactId);
    mds.addMiningAttribute(itemIndex);
    mds.addMiningAttribute(itemId);
    MiningInputStream inputData = new MiningCsvStream( "data/csv/sessions.txt", mds );
    inputData.open();
    MiningDataSpecification metaData = inputData.getMetaData();

    // Get transactional attributes:
    CategoricalAttribute categoryItemId = (CategoricalAttribute)metaData.getMiningAttribute( "itemId" );
    CategoricalAttribute categoryTransactId = (CategoricalAttribute)metaData.getMiningAttribute( "transactionId" );
    NumericAttribute categoryItemIndex  = (NumericAttribute)metaData.getMiningAttribute( "itemIndex" );

    // Create MiningSettings object and assign metadata:
    SequentialSettings miningSettings = new SequentialSettings();
    miningSettings.setDataSpecification( metaData );

    // Assign settings:
    miningSettings.setItemId( categoryItemId );
    miningSettings.setTransactionId( categoryTransactId );
    miningSettings.setItemIndex( categoryItemIndex );
    miningSettings.setMinimumSupport(0.5);
    miningSettings.setGenerateRules(true);
    miningSettings.setMinimumConfidence(0.3);
    miningSettings.verifySettings();

    // Get default mining algorithm specification from 'algorithms.xml':
    MiningAlgorithmSpecification miningAlgorithmSpecification =
      MiningAlgorithmSpecification.getMiningAlgorithmSpecification( "SequentialSimple", null );
    if( miningAlgorithmSpecification == null )
      throw new MiningException( "Can't find application SequentialSimple." );

    // Get class name from algorithms specification:
    String className = miningAlgorithmSpecification.getClassname();
    if( className == null )
      throw new MiningException( "classname attribute expected." );

    // Set and display mining algorithm specification parameters:
    miningAlgorithmSpecification.setMAPValue("minimumItemSize", "1");
    miningAlgorithmSpecification.setMAPValue("maximumItemSize", "-1");
    GeneralUtils.displayMiningAlgSpecParameters(miningAlgorithmSpecification);

    // Create automation parameter, if automation is required:
    MiningAutomationAssignment maa = new MiningAutomationAssignment();
    SequenceNumberAssessment sna = new SequenceNumberAssessment();
    sna.setRuleType(1);  // number of large sequences
    maa.setMiningModelAssessment( sna );
    maa.setMiningAutomationCallback( new SequenceNumberCallback() );
    maa.setMinAssessment(90 );
    maa.setMaxAssessment( 120 );

    // Create algorithm object with default values:
    SequentialAlgorithm algorithm = (SequentialAlgorithm)
        GeneralUtils.createMiningAlgorithmInstance(className);

    // Put it all together:
    algorithm.setMiningInputStream( inputData );
    algorithm.setMiningSettings( miningSettings );
    algorithm.setMiningAlgorithmSpecification( miningAlgorithmSpecification );
    algorithm.setMiningAutomationAssignment( maa );
    // Parameter specific for SequentialAlgorithm but not in MAS:
    algorithm.setExportTransactIds(false);
    algorithm.setExportTransactItemNames( SequentialMiningModel.EXPORT_PMML_NAME_TYPE_XELOPES );
    algorithm.verify();

    // Build the mining model:
    MiningModel model = algorithm.buildModelWithAutomation();
    System.out.println("calculation time [s]: " + algorithm.getTimeSpentToBuildModel());

    // Show results:
    showRules((SequentialMiningModel) model);

    // Write to PMML:
    FileWriter writer = new FileWriter("data/pmml/SequenceModel.xml");
    model.writePmml(writer);

    // Show in browser:
    if (debug == 2) PmmlUtils.openPmmlBrowser("SequenceModel.xml");
  }

  /**
   * Example of building a sequence analysis model.
   *
   * @param args arguments (ignored)
   */
  public static void main(String[] args) {

    try {
      new SequentialBuild().runExample();
    }
    catch (Exception ex) {
      ex.printStackTrace();
    }
  }

  /**
   * Show sequences.
   *
   * @param seqModel model of sequential analysis
   * @exception MiningException cannot show rules
   */
  public static void showRules(SequentialMiningModel seqModel)
   throws MiningException {

      // Get all sequences and rules from model:
      Vector seq   = seqModel.getSequentialRules();
      Vector rules = seqModel.getSequenceRules();

      // Get item and transaction attributes:
      CategoricalAttribute itemId     = (CategoricalAttribute)( (SequentialSettings) seqModel.getMiningSettings() ).getItemId();
      // Get number of sequences and transactions:
      int nSeq            = seq.size();
      int nRules          = 0;
      if (rules != null) nRules = rules.size();
      int transactsNumber = seqModel.getNumberOfTransactions();

      // Show all sequential rules if exist:
      System.out.println();
      System.out.println("Number of sequential rules found: " + nRules);
      for (int i = 0; i < nRules; i++) {
        // New rule:
        System.out.print(i + ": ");

        // Get and show rule:
        RuleSetSeq rss = (RuleSetSeq) rules.elementAt(i);
        int itemSize   = rss.getSize();

        // Premise part of rule:
        ItemSetSeq iss = rss.getPremise();
        int nprem      = rss.getPremise().getSize();
        for (int j = 0; j < nprem; j++) {
          int pN        = iss.getItemAt(j);
          Category cat  = (Category) itemId.getCategory(pN);
          System.out.print(cat.getValue() + " ");
        };
        System.out.print("=> ");

        // Conclusion part of rule:
        for (int j = nprem; j < itemSize; j++) {
          int pN        = rss.getConclusion().getItemAt(j-nprem);
          Category cat  = (Category) itemId.getCategory(pN);
          System.out.print(cat.getValue() + " ");
        }

        // Show support and confidence of rule:
        double Support    = rss.getSupport() * 100.0;
        double Confidence = rss.getConfidence() * 100.0;
        System.out.print("Supp = " + Math.round(Support*100)/100.0 + ", ");
        System.out.print("Conf = " + Math.round(Confidence*100)/100.0 + ", ");

        // Additional measures:
        seqModel.buildLargeSequences();
        double Coverage = seqModel.coverage(rss) * 100.0;
        double Lift     = seqModel.lift(rss);
        System.out.print("Cov = " + Math.round(Coverage*100)/100.0 + ", ");
        System.out.println("Lift = " + Math.round(Lift*100)/100.0);
      };

      // Show all sequences:
      System.out.println();
      System.out.println("Number of sequences found: " + nSeq);
      for (int i = 0; i < nSeq; i++) {
        // New sequence:
        System.out.print(i + ": ");

        // Get sequence:
        ItemSetSeq iss = (ItemSetSeq) seq.elementAt(i);
        int itemSize   = iss.getSize();
        for (int j = 0; j < itemSize; j++) {
          int pN        = iss.getItemAt(j);
          Category cat  = (Category) itemId.getCategory(pN);
          System.out.print(cat.getValue() + " ");
        };

        // Show support of sequence:
        double Support = 100.0 * ((double) iss.getSupportCount()) /
                                 transactsNumber;
        System.out.println(" Supp = " + Math.round(Support*100)/100.0);
      }
  }
}