flatrulesselectionbuild.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.Core.CategoricalAttribute;
import com.prudsys.pdm.Core.Category;
import com.prudsys.pdm.Core.CategoryHierarchy;
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.Multidimensional.MultidimensionalStream;
import com.prudsys.pdm.Input.Multidimensional.OrderAttribute;
import com.prudsys.pdm.Input.Multidimensional.SelectAttribute;
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.Algorithms.FlatRules.FlatRulesAlgorithm;
import com.prudsys.pdm.Utils.GeneralUtils;
import com.prudsys.pdm.Utils.PmmlUtils;

/**
 * Builds flat rule model on taxonomy data and runs selection on rules.
 */
public class FlatRulesSelectionBuild extends BasisExample {

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

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

    // ----------- Open data source as multidimensional stream --------------
    // 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 inputData0 = new MiningCsvStream( "data/csv/sessions.txt", mds );
    inputData0.open();

    // Open multidimensional stream, use only itemId attribute for selection:
    MultidimensionalStream inputData = new MultidimensionalStream(inputData0);
    Vector<String> selectionAttributes = new Vector<String>();
    selectionAttributes.addElement( itemId.getName() );
    inputData.readMultidimensionalStreamData();

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

    // Add taxonomy of items to categoryItemId (for 'sessions.txt'):
    AssociationRulesTaxonomyBuild.addTaxonomy(categoryItemId);

    // -------------------- Apply selection to stream -----------------------
    // Apply selection of all product items to input stream:
    CategoryHierarchy cah = categoryItemId.getTaxonomy();
    Category productRoot  = new Category("2.Produkte");
    Vector products       = cah.getAllChildren(productRoot);
    SelectAttribute sa = new SelectAttribute( categoryItemId.getName(), productRoot.getDisplayValue() );
    for (int i = 0; i < products.size(); i++) {
      Category product = (Category) products.elementAt(i);
      sa.addCategory( product.getDisplayValue() );
    };
    SelectAttribute[] selArr = {sa};
    inputData.runSelections(selArr);
    inputData.reset();

    // ------------------------- Build flat 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( 0.01 );
    miningSettings.verifySettings();

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

    // 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", "5");
    miningAlgorithmSpecification.setMAPValue("rulesOrderType", String.valueOf(FlatRulesAlgorithm.ORDER_LIFT) );
    miningAlgorithmSpecification.setMAPValue("createLargeItemSets", "true");
    GeneralUtils.displayMiningAlgSpecParameters(miningAlgorithmSpecification);

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

    // ------------------- Show and export flat rules ----------------------
    // Show results:
    FlatRulesBuild.showRules((AssociationRulesMiningModel) model);

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

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

    // --------------- Perform selection and ordering of rules --------------
    runRuleSelection( (AssociationRulesMiningModel) model );
  }

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

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

  /**
   * Performs selection and ordering rules.
   *
   * @param ruleModel flat rules model
   * @exception MiningException mining exception
   */
  public static void runRuleSelection(AssociationRulesMiningModel ruleModel)
    throws MiningException {

    // --------------------- Convert rules as stream ------------------------
    // Convert rules into mining input stream:
    ruleModel.setMisExportType(AssociationRulesMiningModel.EXPORT_MIS_ASSOCIATION_RULES_FLAT );
    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);

    // ----------------- Create multidimenional stream ----------------------
    mis.reset();
    MultidimensionalStream multiRules = new MultidimensionalStream(mis);
    multiRules.readMultidimensionalStreamData();

    // ------------- Create selection and ordering conditions ---------------
    CategoricalAttribute itemId = (CategoricalAttribute)( (AssociationRulesSettings) ruleModel.getMiningSettings() ).getItemId();
    CategoryHierarchy cah = itemId.getTaxonomy();

    // Select all software products in premise:
    Category softwareRoot = new Category("2.1.Software");
    Vector software       = cah.getAllChildren(softwareRoot);
    SelectAttribute sa    = new SelectAttribute( "itemIdPremise", softwareRoot.getDisplayValue() );
    for (int i = 0; i < software.size(); i++) {
      Category soft = (Category) software.elementAt(i);
      sa.addCategory( soft.getDisplayValue() );
    };

    // Select all algorithm and component products in conclusion:
    SelectAttribute sa2   = new SelectAttribute( "itemIdConclusion", "2.3.Algorithmen");
    sa2.addCategory("2.2.Komponenten");

    SelectAttribute[] selArr = {sa, sa2};

    // Order by 1. itemIdPremise, 2. support (increasing), 3. lift (decreasing):
    OrderAttribute oa1 = new OrderAttribute( "itemIdPremise", OrderAttribute.UP );
    OrderAttribute oa2 = new OrderAttribute( "support", OrderAttribute.UP );
    OrderAttribute oa3 = new OrderAttribute( "lift", OrderAttribute.DOWN );
    OrderAttribute[] ordArr = {oa1, oa2, oa3};

    // ------------- Apply selection and ordering to stream -----------------
    // Run selection and ordering:
    multiRules.setOrdering(ordArr);
    multiRules.runSelections(selArr);
    System.out.println("Selected rules as mining input stream: " + multiRules);
  }

}