wekatools.java

一个很好的LIBSVM的JAVA源码。对于要研究和改进SVM算法的学者。可以参考。来自数据挖掘工具YALE工具包。

/*
 *  YALE - Yet Another Learning Environment
 *  Copyright (C) 2001-2004
 *      Simon Fischer, Ralf Klinkenberg, Ingo Mierswa, 
 *          Katharina Morik, Oliver Ritthoff
 *      Artificial Intelligence Unit
 *      Computer Science Department
 *      University of Dortmund
 *      44221 Dortmund,  Germany
 *  email: [e-mail]
 *  web:   http://yale.cs.uni-dortmund.de/
 *
 *  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., 59 Temple Place, Suite 330, Boston, MA 02111-1307
 *  USA.
 */
package edu.udo.cs.yale.tools;

import edu.udo.cs.yale.example.Attribute;
import edu.udo.cs.yale.example.Example;
import edu.udo.cs.yale.example.ExampleSet;
import edu.udo.cs.yale.example.MemoryExampleTable;
import edu.udo.cs.yale.example.ListDataRowReader;
import edu.udo.cs.yale.example.DoubleArrayDataRow;
import edu.udo.cs.yale.example.ExampleReader;

import weka.core.FastVector;
import weka.core.Instance;
import weka.core.Instances;

import java.util.List;
import java.util.LinkedList;
import java.util.Iterator;
import java.util.Enumeration;
import java.util.jar.JarFile;
import java.util.jar.JarEntry;
import java.io.IOException;

/** This class contains static methods for converting 
 *  <a href="http://www.cs.waikato.ac.nz/~ml/weka/">Weka</a> Instances
 *  to Yale ExampleSet and vice versa.
 *
 *  @version $Id: WekaTools.java,v 2.18 2004/09/01 16:04:46 ingomierswa Exp $
 */
public class WekaTools {

    /** Invokes toYaleExampleSet(instances, null). */
    public static ExampleSet toYaleExampleSet(Instances instances) {
	return toYaleExampleSet(instances, null);
    }

    /** Creates a Yale example set from Weka instances. Only a label can be used as special attributes, other
     *  types of special attributes are not supported. If attributeNames is not null, the string is used as 
     *  prefix plus a number as attribute names. */
    public static ExampleSet toYaleExampleSet(Instances instances, String attributeNames) {
	int classIndex = instances.classIndex();

	// create example table
	List attributes = new LinkedList();
	int number = 1; // use for attribute names
	for (int i = 0; i < instances.numAttributes(); i++) {
	    weka.core.Attribute wekaAttribute = instances.attribute(i);
	    Attribute attribute = new Attribute(wekaAttribute.name(), 
						(wekaAttribute.isNominal() ? Ontology.NOMINAL : Ontology.REAL),
						Ontology.SINGLE_VALUE,
						Attribute.UNDEFINED_BLOCK_NR,
						null);
	    if ((i != classIndex) && (attributeNames != null)) {
		attribute.setName(attributeNames + "_" + (number++));
	    }
	    attributes.add(attribute);
	}
	
	Attribute label = (Attribute)attributes.get(classIndex);
	label.setName("label");
	
	MemoryExampleTable table = new MemoryExampleTable(attributes);
	
	// create data
	List dataList = new LinkedList();
	int numberOfYaleAttributes = instances.numInstances() - ((classIndex >= 0) ? 1 : 0);
	for (int i = 0; i < instances.numInstances(); i++) {
	    Instance instance = instances.instance(i);
	    double[] data = new double[numberOfYaleAttributes];
	    for (int a = 0; a < instances.numAttributes(); a++) {
		double wekaValue = instance.value(a);
		if (table.getAttribute(a).isNominal()) {
		    String nominalValue = instances.attribute(a).value((int)wekaValue);
		    data[a] = table.getAttribute(a).mapString(nominalValue);
		} else {
		    data[a] = wekaValue;
		}
	    }
	    dataList.add(new DoubleArrayDataRow(data));
	}
	
	// handle label extra
	table.readExamples(new ListDataRowReader(dataList.iterator()));
	
	// create and return example set
	return table.createCompleteExampleSet(label, null, null, null);
    }

    /** Gets an example and creates a Weka instance. */
    public static Instance toWekaInstance(Example example, 
					  Instances instances, 
					  Attribute label,
					  boolean labelKnown) {
	Instance instance = new Instance(example.getNumberOfAttributes() + (label != null ? 1 : 0));
	instance.setDataset(instances);

	if (example.getWeightAttribute() != null)
		instance.setWeight(example.getWeight());
		
	for (int a = 0; a < example.getNumberOfAttributes(); a++) {
	    double v = example.getValue(a);
	    if (Double.isNaN(v)) {
		instance.setMissing(a);
	    } else {
		if (example.getAttribute(a).isNominal()) {
		    try {
			instance.setValue(a, example.getValueAsString(a));
		    } catch (Exception e) {
			throw new RuntimeException("Cannot set attribute value: " + e);
		    }
		} else {
		    instance.setValue(a, v);
		}
	    }
	}

	if (label != null) {
	    if (!labelKnown) instance.setClassMissing();
	    else {
		double v = example.getValue(label);
		if (Double.isNaN(v)) {
		    instance.setClassMissing();
		} else {
		    if (Ontology.ATTRIBUTE_VALUE_TYPE.isA(label.getValueType(), 
							  Ontology.NOMINAL)) {
			try {
			    instance.setClassValue(example.getAsString(label, v));
			} catch (Exception e) {
			    throw new RuntimeException("Cannot set class value: " + e);
			}
		    } else {
			try {
			    instance.setClassValue(v);
			} catch (Exception e) {
			    throw new RuntimeException("Cannot set class value: " + e);
			}
		    }
		}
	    }
	}
	return instance;
    }


    /** Gets an ExampleSet and creates Weka instances. The label attribute has must be the true label (for
     *  regular learning), the predicted label (for prediction with learned models) or null (for association
     *  learning, clustering, etc.) The parameter labelKnown indicates if the label is known (e.g. for learning)
     *  or missing (e.g. for prediction). */
    public static Instances toWekaInstances(ExampleSet exampleSet, 
					    String name, 
					    Attribute label, 
					    boolean labelKnown) {
	FastVector attributeVector = new FastVector(exampleSet.getNumberOfAttributes());
	for (int i = 0; i < exampleSet.getNumberOfAttributes(); i++) {
	    attributeVector.addElement(toWekaAttribute(exampleSet.getAttribute(i)));
	}
	
	if (label != null) attributeVector.addElement(toWekaAttribute(label));
	Instances instances = new Instances(name, attributeVector, exampleSet.getSize());

	if (label != null) instances.setClassIndex(attributeVector.size()-1);
	else instances.setClassIndex(-1);

	ExampleReader r = exampleSet.getExampleReader();
	while (r.hasNext()) {
	    instances.add(toWekaInstance(r.next(), instances, label, labelKnown));
	}
	return instances;
    }

    /** Converts an {@link Attribute} to a Weka attribute. */
    public static weka.core.Attribute toWekaAttribute(Attribute attribute) {
	if (attribute == null) return null;
	if (Ontology.ATTRIBUTE_VALUE_TYPE.isA(attribute.getValueType(), Ontology.NOMINAL)) {
	    FastVector nominalValues = new FastVector(attribute.getValues().size());
	    Iterator c = attribute.getValues().iterator();
	    while (c.hasNext()) {
		nominalValues.addElement((String)c.next()); 
	    }
	    return new weka.core.Attribute(attribute.getName(), nominalValues);
	} else {
	    return new weka.core.Attribute(attribute.getName());
	}
    }


    public static JarFile getWekaJar() throws IOException {
	return new JarFile(ParameterService.getLibraryFile("weka.jar"));
    }

    public static String[] getWekaClasses(Class superclass) {
	return getWekaClasses(superclass, null, true);
    }

    public static String[] getWekaClasses(Class superclass, String packageName, boolean includePackage) {
	JarFile jar = null;
	try {
	    jar = getWekaJar();
	} catch (IOException e) {
	    return new String[0];
	}
	List classes = new LinkedList();
	Tools.findImplementationsInJar(jar, superclass, classes);
	if (packageName != null) {
	    Iterator i = classes.iterator();
	    while (i.hasNext()) {
		if (includePackage && (((String)i.next()).indexOf(packageName) == -1)) i.remove();
		if (!includePackage && (((String)i.next()).indexOf(packageName) != -1)) i.remove();
	    }
	}
	String[] names = new String[classes.size()];
	classes.toArray(names);
	return names;
    }

    public static String[] getWekaParameters(List yaleParameters) {
	String[] parameters = new String[yaleParameters.size()*2];
	Iterator i = yaleParameters.iterator();
	int j = 0;
	while (i.hasNext()) {
	    Object[] parameter = (Object[])i.next();
	    parameters[j++] = "-"+(String)parameter[0];
	    parameters[j++] = (String)parameter[1];	    
	}
	return parameters;
    }

}