bnctester.java

bayes network classifier toolbox 贝叶斯网络分类工具箱

/**
 *  JBNC - Bayesian Network Classifiers Toolbox <p>
 *
 *  Latest release available at http://sourceforge.net/projects/jbnc/ <p>
 *
 *  Copyright (C) 1999-2003 Jarek Sacha <p>
 *
 *  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. <p>
 *
 *  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. <p>
 *
 *  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., 59 Temple
 *  Place - Suite 330, Boston, MA 02111-1307, USA. <br>
 *  http://www.fsf.org/licenses/gpl.txt
 */
package jbnc.util;

import BayesianNetworks.BayesNet;
import BayesianNetworks.ProbabilityFunction;
import InferenceGraphs.InferenceGraph;
import InferenceGraphs.InferenceGraphNode;
import QuasiBayesianInferences.QBInference;
import gnu.getopt.Getopt;
import jbnc.dataset.AttributeType;
import jbnc.dataset.Dataset;
import jbnc.dataset.DatasetReader;
import jbnc.dataset.NamesReader;

import java.text.DecimalFormat;
import java.text.NumberFormat;
import java.util.HashMap;
import java.util.Vector;

/**
 *  Description of the Class
 *
 * @author     Jarek Sacha
 * @since
 */
public class BNCTester {

    /*
     *  Private and protected constants
     */
    protected final static String classNameDefault = "class";

    /*
     *  Private and protected member variables
     */
    private final static String programName = "BNCTester";
    private final static String programVersion = "v.0.02";

    // Command line configuration parameters
    protected boolean useTimer = false;
    protected boolean debugMode = false;
    protected String dataFileName = null;
    protected String namesFileName = null;
    protected String netFileName = null;
    protected String className = null;

    // Test data set
    protected Dataset dataset;

    // Network representation
    protected InferenceGraph graph;
    // Graph representing the network
    protected InferenceGraphNode[] attribNodes;
    // References to graph nodes, order as in cases
    protected InferenceGraphNode classNode;
    // Reference to class node in the graph

    // Classification statistics
    protected int nbPass;
    protected int nbFail;


    /**
     * @param  argv  Description of Parameter
     */
    public static void main(String argv[]) {
        String usageMsg = "\n"
                + "USAGE: BNCTester <options> network_file [class_name]\n"
                + "\n"
                + "where:\n"
                + "  -d              Print debugging information.\n"
                + "  -f <file_stem>  Load test data in C4.5 format (.names + .test),\n"
                + "                  file_stem.names - file with specification of attributes,\n"
                + "                  file_stem.test  - file with test cases.\n"
                + "  -h <test_data>  Load test data in comma delimited format. First row contains\n"
                + "                  attribute names.\n"
                + "  -t              Print execution time."
                + "\n"
                + "  network_file    Classifier network definition file.\n"
                + "  class_name      Name of the class variable. The default value is 'class'.\n"
                + "\n"
                + "EXAMPLE: BNCTester -tf monk1 monk1.bif";

        try {
            Timer t = new Timer();

            // Print header
            System.out.println("\nTester for Bayesian network classifiers.\n");

            BNCTester tester = new BNCTester();

            // Process command line arguments
            if (!tester.processCommandLine(argv)) {
                System.out.println(usageMsg);
                return;
            }

            // Run the test
            tester.run();

            // Print execution time
            if (tester.useTimer) {
                t.println("\nExecution time = ");
            }
        } catch (Exception e) {
            e.printStackTrace();
        }
    }


    /**
     *  Sets the DebugMode attribute of the BNCTester object
     *
     * @param  mode  The new DebugMode value
     */
    public void setDebugMode(boolean mode) {
        debugMode = mode;
    }


    /**
     * @param  net            Description of Parameter
     * @param  dataset        Description of Parameter
     * @return                Confusion matrix.
     * @exception  Exception  Description of Exception
     */
    public Result test(BayesNet net, Dataset dataset) throws Exception {
        clear();

        this.graph = null;
        this.dataset = dataset;
        this.className = null;

//    verifyDatasetVsNetwork();

        testClassifier(net);

        if (debugMode) {
            printClassificationReport();
        }

        return new Result(nbPass, nbFail);
    }


    /**
     * @param  net            Description of Parameter
     * @param  dataset        Description of Parameter
     * @return                Confusion matrix.
     * @exception  Exception  Description of Exception
     */
    public Result test_old(BayesNet net, Dataset dataset) throws Exception {
        return test_old(new InferenceGraph(net), dataset);
    }


    /**
     * @param  graph          Description of Parameter
     * @param  dataset        Description of Parameter
     * @return                Confusion matrix.
     * @exception  Exception  Description of Exception
     */
    public Result test_old(InferenceGraph graph, Dataset dataset) throws Exception {
        clear();

        this.graph = graph;
        this.dataset = dataset;
        this.className = dataset.names[dataset.names.length - 1].getName();

        verifyDatasetVsNetwork();

        testClassifier_old();

        if (debugMode) {
            printClassificationReport();
        }

        return new Result(nbPass, nbFail);
    }


    /**  Print test results to the standard output. */
    public void report() {
        printClassificationReport();
    }


    /**
     *  Print test results to the standard output in the short form (single line).
     *
     * @exception  Exception  Description of Exception
     */
    public void reportShort() throws Exception {
        Result r = new Result(nbPass, nbFail);
        r.reportShort();
    }


    /**
     *  Process command line arguments.
     *
     * @param  argv  Description of Parameter
     * @return       'false' when command line is invalid.
     */
    protected boolean processCommandLine(String[] argv) {

        try {
            Getopt g = new Getopt("ERROR", argv, "df:h:t");

            int c;
            String arg = null;
            boolean is_f = false;
            boolean is_h = false;
            while ((c = g.getopt()) != -1) {
                switch (c) {
                    case 'd':
                        debugMode = true;
                        break;
                    case 'f':
                        is_f = true;
                        arg = g.getOptarg();
                        dataFileName = arg + ".test";
                        namesFileName = arg + ".names";
                        break;
                    case 'h':
                        is_h = true;
                        dataFileName = g.getOptarg();
                        break;
                    case 't':
                        useTimer = true;
                        break;
                    case '?':
                        break;
                    default:
                        System.out.println("\nERROR: getopt() returned: " + c);
                        return false;
                }
            }

            if (is_f && is_h) {
                System.out.println("\nERROR: Options 'f' and 'h' cannot appear together.");
                return false;
            }

            if (!is_f && !is_h) {
                System.out.println("\nERROR: Either option 'f' or 'h' has to be specified.");
                return false;
            }

            // Classifier network file name
            int index = g.getOptind();
            if (index >= argv.length) {
                System.out.println("\nERROR: Classifier network file not specified.");
                return false;
            }
            netFileName = argv[index];

            // Class name
            ++index;
            if (index < argv.length) {
                className = argv[index];
            } else {
                className = classNameDefault;
            }

            if (debugMode) {
                System.out.println("debug mode   = " + debugMode);
                System.out.println("use timer    = " + useTimer);
                System.out.println("data file    = " + dataFileName);
                System.out.println("names file   = " + namesFileName);
                System.out.println("network file = " + netFileName);
                System.out.println("class name   = " + className);
            }

            ++index;
            if (index >= argv.length) {
                return true;
            }

            // Staff left on the command line
            System.out.print("\nERROR: Too many arguments in the command line. Extraneous: ");
            for (; index < argv.length; ++index) {
                System.out.print(argv[index] + ", ");
            }

            System.out.println("");
        } catch (Exception e) {
            System.out.println("\nError: " + e);
            if (debugMode) {
                e.printStackTrace();
            }
        }

        return false;
    }


    /*
     *
     */
    protected void run() {
        try {
            Timer tLoadNet = new Timer();
            loadNetwork();
            tLoadNet.stop();

            Timer tLoadData = new Timer();
            loadDataset();
            tLoadData.stop();

//      Timer tTest = new Timer();
//      test(graph, dataset);
//      tTest.stop();

            Timer tVerify = new Timer();
            verifyDatasetVsNetwork();
            tVerify.stop();

            Timer tClassify = new Timer();
            testClassifier_old();
//      testClassifier();
            tClassify.stop();

            Timer tReport = new Timer();
            printClassificationReport();
            tReport.stop();

            if (useTimer) {
                tLoadNet.println("\nLoad net  = ");
                tLoadData.println("Load data = ");
//        tTest.println(     "Test      = ");
                tVerify.println("Verify    = ");
                tClassify.println("Classify  = ");
                tReport.println("Report    = ");
            }
        } catch (Exception e) {
            System.out.println("\nError: " + e);
            if (debugMode) {
                e.printStackTrace();
            }
        }
    }


    /*
     *
     */
    protected void loadNetwork() throws Exception {
        if (debugMode) {
            System.out.println("\nLoading network: " + netFileName);
        }

        if (netFileName == null) {
            throw new Exception("Name of the network file is not set.");
        }

        // Load network
        graph = new InferenceGraph(netFileName);

        if (debugMode) {
            Vector graphNodes = graph.get_nodes();