associationrulesolapbuild.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 javax.olap.OLAPException;

import com.prudsys.pdm.Automat.MiningAutomationAssignment;
import com.prudsys.pdm.Core.CategoricalAttribute;
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.AssociationRules.AssociationRulesAlgorithm;
import com.prudsys.pdm.Models.AssociationRules.AssociationRulesMiningModel;
import com.prudsys.pdm.Models.AssociationRules.AssociationRulesSettings;
import com.prudsys.pdm.Models.AssociationRules.RulesNumberAssessment;
import com.prudsys.pdm.Models.AssociationRules.RulesNumberCallback;
import com.prudsys.pdm.Olap.OlapEngine;
import com.prudsys.pdm.Olap.Cursor.CubeCursor;
import com.prudsys.pdm.Olap.Cursor.DimensionCursor;
import com.prudsys.pdm.Olap.Cursor.EdgeCursor;
import com.prudsys.pdm.Olap.Cursor.RowDataMetaData;
import com.prudsys.pdm.Olap.Metadata.Cube;
import com.prudsys.pdm.Olap.Metadata.CubeClass;
import com.prudsys.pdm.Olap.Metadata.Dimension;
import com.prudsys.pdm.Olap.Metadata.DimensionClass;
import com.prudsys.pdm.Olap.Metadata.Measure;
import com.prudsys.pdm.Olap.Metadata.MeasureClass;
import com.prudsys.pdm.Olap.Metadata.Schema;
import com.prudsys.pdm.Olap.Metadata.SchemaClass;
import com.prudsys.pdm.Olap.Metadata.Measures.AggregationMeasure;
import com.prudsys.pdm.Olap.Query.Core.CubeView;
import com.prudsys.pdm.Olap.Query.Core.DimensionView;
import com.prudsys.pdm.Olap.Query.Core.EdgeView;
import com.prudsys.pdm.Olap.Query.Core.MeasureView;
import com.prudsys.pdm.Utils.GeneralUtils;
import com.prudsys.pdm.Utils.PmmlUtils;

/**
 * Builds an association rule model and applies OLAP analysis
 * to the results.
 */
public class AssociationRulesOlapBuild extends BasisExample {

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

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

    // --------------------- Open data source  ------------------------------
    // 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();

    // Get meta data:
    MiningDataSpecification metaData        = inputData.getMetaData();
    CategoricalAttribute categoryItemId     = (CategoricalAttribute)metaData.getMiningAttribute( "itemId" );
    CategoricalAttribute categoryTransactId = (CategoricalAttribute)metaData.getMiningAttribute( "transactionId" );

    // ---------------------- Build association rules -----------------------
    // Create MiningSettings object and assign metadata:
    AssociationRulesSettings miningSettings = new AssociationRulesSettings();
    miningSettings.setDataSpecification( metaData );

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

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

    // 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();
    maa.setMiningModelAssessment( new RulesNumberAssessment() );
    maa.setMiningAutomationCallback( new RulesNumberCallback() );
    maa.setMinAssessment(4000);
    maa.setMaxAssessment(5000);
    maa.setMaxIterationNumber(30);

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

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

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

    // ----------------- Show and export association rules ------------------
    // Show results:
    if (debug == 1) AssociationRulesBuild.showRules((AssociationRulesMiningModel) model);

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

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

    // ------------------ Perform OLAP analysis of rules --------------------
    runOlapAnalysis( (AssociationRulesMiningModel) model );
  }

  /**
   * Example of building an association rules model with subsequent
   * OLAP analysis.
   *
   * @param args arguments (ignored)
   */
  public static void main(String[] args) {

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

  /**
   * Performs OLAP analysis on association rules.
   *
   * @param ruleModel association rules model
   * @exception MiningException mining exception
   * @exception OLAPException OLAP exception
   */
  public static void runOlapAnalysis(AssociationRulesMiningModel ruleModel)
    throws MiningException, OLAPException {

    // --------------------- Convert rules as stream ------------------------
    // Convert rules into mining input stream:
    ruleModel.setMisExportType(AssociationRulesMiningModel.EXPORT_MIS_ASSOCIATION_RULES );
    ruleModel.setMisExportCharType( AssociationRulesMiningModel.EXPORT_MISCHAR_SUPPORT +
                                    AssociationRulesMiningModel.EXPORT_MISCHAR_LIFT);
    MiningInputStream mis = ruleModel.toMiningInputStream();
    System.out.println();
//    System.out.println("Rules as mining input stream: " + mis);
    mis.reset();

    // -------------------- Create OLAP Schema with Cube ----------------------
    // Create local OLAP engine:
    System.out.println("Create OLAP schema...");
    OlapEngine olap = new OlapEngine();

    // Assign source stream to OLAP engine:
    olap.setInputStream(mis);

    // Create flat dimensions:
    DimensionClass dimClass = (DimensionClass) olap.getDimension();
    Dimension itemID   = dimClass.createDimension("ItemID", "itemId");
    Dimension concFlag = dimClass.createDimension("ConclusionFlag", "conclusionFlag");

    // Create measures:
    MeasureClass measClass = (MeasureClass) olap.getMeasure();
    Measure itemCount = measClass.createAggregationMeasure("ItemCount",
      AggregationMeasure.COUNT, olap.getInputAttribute("itemId") );
    Measure meanSupp  = measClass.createAggregationMeasure("MeanSupport",
      AggregationMeasure.MEAN, olap.getInputAttribute("support") );
    Measure maxSupp   = measClass.createAggregationMeasure("MaxSupport",
      AggregationMeasure.MAX, olap.getInputAttribute("support") );
    Measure meanLift  = measClass.createAggregationMeasure("MeanLift",
      AggregationMeasure.MEAN, olap.getInputAttribute("lift") );
    Measure maxLift   = measClass.createAggregationMeasure("MaxLift",
      AggregationMeasure.MAX, olap.getInputAttribute("lift") );

    // Create measure dimension:
    Dimension meas    = dimClass.createMeasureDimension("Measures");
    meas.addInputAttribute(itemCount);
    meas.addInputAttribute(meanSupp);
    meas.addInputAttribute(maxSupp);
    meas.addInputAttribute(meanLift);
    meas.addInputAttribute(maxLift);

    // Create an overall holding schema:
    SchemaClass schemaClass  = (SchemaClass) olap.getSchema();
    Schema dimensionalSchema = schemaClass.createSchema("Association Rules OLAP");
    dimensionalSchema.addDimension(itemID);
    dimensionalSchema.addDimension(concFlag);
    dimensionalSchema.addDimension(meas);

    // Create the cube:
    CubeClass cubeClass = (CubeClass) olap.getCube();
    Cube ruleCube = cubeClass.createCube("Rule Cube");

    // Create the cube dimension associations:
    ruleCube.addDimension(itemID);
    ruleCube.addDimension(concFlag);
    ruleCube.addDimension(meas);

    // Add cube to the schema:
    dimensionalSchema.addCube(ruleCube);

    // Add schema to OLAP engine:
    olap.setCurrentSchema(dimensionalSchema);

    // Init OLAP task:
    olap.init();

    System.out.println("schema: " + dimensionalSchema);

    // -------------------- Run query against the cube ----------------------
    // Create a dimension view for each dimension:
    System.out.println("Create OLAP query...");
    DimensionView itemView = (DimensionView) olap.createDimensionView();
    itemView.setDimension(itemID);
    DimensionView concView = (DimensionView) olap.createDimensionView();
    concView.setDimension(concFlag);
    MeasureView measView = (MeasureView) olap.createMeasureView();
    measView.setDimension(meas);

    // Create the query cube view and add edges and the measure:
    CubeView query = (CubeView) olap.createCubeView(ruleCube);

    // Create a columns edge and add item views:
    EdgeView columns = (EdgeView) query.createOrdinateEdge();
    columns.addDimensionView(itemView);

    // Create a rows edge and add the conclusion view
    EdgeView rows = (EdgeView) query.createOrdinateEdge();
    rows.addDimensionView(concView);

    // Create a pages edge and add measure dimension view:
    EdgeView pages = (EdgeView) query.createPageEdge();
    pages.addDimensionView(measView);

    // Create the cube cursor associated to the query:
    CubeCursor dataCursor = (CubeCursor) query.createCursor();
    System.out.println("aggregateStream: " + dataCursor.getResultTable());

    // Get page and both ordinate cursors:
    EdgeCursor pageCursor = (EdgeCursor) dataCursor.getPageEdge().iterator().next();
    EdgeCursor columnCursor = (EdgeCursor) dataCursor.getOrdinateEdge().get(0);
    EdgeCursor rowCursor = (EdgeCursor) dataCursor.getOrdinateEdge().get(1);

    // Get dimension cursors of all page and ordinate cursors:
    DimensionCursor measureCursor = (DimensionCursor) pageCursor.getDimensionCursor().get(0);
    DimensionCursor itemCursor = (DimensionCursor) columnCursor.getDimensionCursor().get(0);
    DimensionCursor concCursor = (DimensionCursor) rowCursor.getDimensionCursor().get(0);

    // ----------------------- Show query result -------------------------
    // Iterate through all dimension cursors:
    System.out.println("Iterate OLAP cursor...");
    RowDataMetaData rmd = (RowDataMetaData) measureCursor.getMetaData();
    itemCursor.beforeFirst();
    while ( itemCursor.next() ) {
      System.out.println( "itemID: " + itemCursor.getString(1) );

      concCursor.beforeFirst();
      while ( concCursor.next() ) {
        System.out.println( "concFlag: " + concCursor.getString(1) );

        measureCursor.beforeFirst();
        while ( measureCursor.next() ) {
          for (int i = 0; i < rmd.getColumnCount(); i++) {
            String mname = rmd.getColumnName(i+1);
            System.out.print(mname + " = " + measureCursor.getDouble(i+1) + " ");
          }
          System.out.println();
        }
      }
      System.out.println();
    }
  }
}