decisiontreeapply.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.FileReader;

import com.prudsys.pdm.Core.CategoricalAttribute;
import com.prudsys.pdm.Core.Category;
import com.prudsys.pdm.Core.MiningDataSpecification;
import com.prudsys.pdm.Core.MiningModel;
import com.prudsys.pdm.Input.MiningFilterStream;
import com.prudsys.pdm.Input.MiningInputStream;
import com.prudsys.pdm.Input.MiningVector;
import com.prudsys.pdm.Input.Records.Arff.MiningArffStream;
import com.prudsys.pdm.Models.Classification.ClassificationRateAssessment;
import com.prudsys.pdm.Models.Classification.GainsChartAssessment;
import com.prudsys.pdm.Models.Classification.DecisionTree.DecisionTreeMiningModel;
import com.prudsys.pdm.Models.Classification.DecisionTree.DecisionTreeNode;
import com.prudsys.pdm.Models.Supervised.SupervisedMiningSettings;

/**
 * Applies a decision tree which is read from the PMML file 'DecisionTreeModel.xml'
 * to a data set.
 */
public class DecisionTreeApply extends BasisExample {

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

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

    // Read decision tree model from PMML file:
    MiningModel model = new DecisionTreeMiningModel();
    FileReader reader = new FileReader("data/pmml/DecisionTreeModel.xml");
    model.readPmml(reader);
    MiningDataSpecification modelMetaData = model.getMiningSettings().getDataSpecification();
    CategoricalAttribute modelTargetAttribute = (CategoricalAttribute)
      ((SupervisedMiningSettings) model.getMiningSettings()).getTarget();
    System.out.println("-------------> PMML model read successfully");

    // Open data source and get metadata:
    MiningInputStream inputData0 = new MiningArffStream( "data/arff/soybeanTest.arff");

    // Transform input data (dynamically) if required:
    MiningInputStream inputData = inputData0;
    if ( modelMetaData.isTransformed() )
      inputData = new MiningFilterStream(inputData0, modelMetaData.getMiningTransformationActivity());

    MiningDataSpecification inputMetaData     = inputData.getMetaData();
    CategoricalAttribute inputTargetAttribute = (CategoricalAttribute)
      inputMetaData.getMiningAttribute( modelTargetAttribute.getName() );

    // Show meta data:
    System.out.println("Prediction:");
    for (int i = 0; i < inputMetaData.getAttributesNumber(); i++) {
      System.out.print(inputMetaData.getMiningAttribute(i).getName() + " ");
    };

    // Show classification results:
    System.out.println();
    int i     = 0;
    int wrong = 0;
    while (inputData.next()) {
      // Make prediction:
      MiningVector vector  = inputData.read();
      DecisionTreeNode dtn = (DecisionTreeNode) model.applyModel(vector);
      double predicted     = dtn.getScore();  // or: predicted = model.applyModelFunction(vector);
      Category predTarCat  = modelTargetAttribute.getCategory(predicted);
      double[] dist        = dtn.getDistribution();

      // Output and stats:
      double realTarCat    = vector.getValue(inputTargetAttribute);
      Category tarCat      = inputTargetAttribute.getCategory(realTarCat);

      System.out.println(" " + ++i +": " + vector + " -> " + predTarCat);
      System.out.print("dist: ");
      for (int j = 0; j < dist.length; j++)
        System.out.print(dist[j] + " ");
      System.out.println();

      if (predTarCat == null || ! predTarCat.equals(tarCat) ) {
        wrong = wrong + 1;
      };
    };
    System.out.println("classification rate = " + (100.0 - ((double) wrong / i)*100.0) );

    // Same classification rate via assessment class:
    inputData.reset();
    ClassificationRateAssessment cra = new ClassificationRateAssessment();
    cra.setMiningModel(model);
    cra.setAssessmentData(inputData);
    System.out.println("classification rate (assessment class) = " + cra.calculateAssessment());

    // Calculate Gains chart for two-class problems:
    if ( modelTargetAttribute.getCategoriesNumber() == 2) {
      inputData.reset();
      GainsChartAssessment gc = new GainsChartAssessment();
      gc.setMiningModel(model);
      gc.setAssessmentData(inputData);
      gc.setUseNormalization(false);
      gc.setUseInterpolation(false);
      double[] gcval = gc.calculateGainsChart();
      System.out.println("G a i n s  C h a r t: ");
      for (int j = 0; j < gcval.length; j++)
        System.out.print(gcval[j]+" ");
      System.out.println();
      inputData.reset();
      System.out.println("Gains chart rate = " + gc.calculateAssessment());
    };
  }

  /**
   * Example of using a decision tree for classification.
   *
   * @param args arguments (ignored)
   */
  public static void main(String[] args) {

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

}