⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 okcsaver.java

📁 MacroWeka扩展了著名数据挖掘工具weka
💻 JAVA
字号:
package chen.macroweka.core.converters;


import weka.core.*;
import chen.macroweka.filters.unsupervised.instance.RemoveMissingValues;
import weka.filters.Filter;

import java.io.*;

/**
 * Implementation of class
 * 
 */
public class OKCSaver extends AbstractSaver
{
    private String m_fileStem;
    private File m_destMetaFile;
    private BufferedWriter m_metaWriter;
    private File m_destDataFile;
    private BufferedWriter m_dataWriter;

    private boolean m_isStructureWritten = false;

    /**
     * Default implementation throws an IOException.
     *
     * @param file the File
     * @throws java.io.IOException always
     */
    public void setDestination(File file) throws IOException
    {
        if ( file == null ) {
            throw new IOException( "Destination file object is null!" );
        }

        String fname = file.getName();
        String fileStem;
        String path = file.getParent();
        if ( path != null ) {
            path += File.separator;
        } else {
            path = "";
        }
        if ( fname.indexOf( '.' ) < 0 ) {
            fileStem = fname;
            fname += ".meta";
        } else {
            fileStem = fname.substring( 0, fname.indexOf( '.' ) );
            fname = fileStem + ".meta";
        }
        m_fileStem = fileStem;
        file = new File( path + fname );

        m_destMetaFile = file;
        try {
            BufferedWriter bw = new BufferedWriter( new FileWriter( m_destMetaFile ) );
            m_metaWriter = bw;
        } catch (FileNotFoundException ex) {
            throw new IOException( "File not found : " + ( path + fname ) );
        }

        m_destDataFile = new File( path + fileStem + ".okc" );
        try {
            BufferedWriter bw = new BufferedWriter( new FileWriter( m_destDataFile ) );
            m_dataWriter = bw;
        } catch (FileNotFoundException ex) {
            throw new IOException( "File not found : " + ( path + fname ) );
        }
    }

    /**
     * Saves the structure (internally the header) of the data set.
     *
     * @param instances the set of Instances
     * @throws java.io.IOException if there is no source or saving fails
     */
    public void saveStructure(Instances structure) throws IOException
    {
        if ( m_destDataFile == null ) {
            throw new IOException( "No destination has been specified" );
        }

        if ( m_metaWriter != null ) {
            writeHeaderToMeta(structure);
        }
        writeHeaderToOKC(structure);

        m_isStructureWritten = true;
    }

    private void writeHeaderToOKC(Instances structure) throws IOException
    {
        m_dataWriter.write(structure.numAttributes() + " " +
                           structure.numInstances() + "\n");
        for (int i=0; i<structure.numAttributes(); ++i) {
            Attribute attr = structure.attribute( i );
            m_dataWriter.write(attr.name() + "\n");
        }
        for (int i=0; i<structure.numAttributes(); ++i) {
            AttributeStats as = structure.attributeStats( i );
            if (as.numericStats != null) {
                m_dataWriter.write(Utils.doubleToString( as.numericStats.min, 3 ) + " " +
                                   Utils.doubleToString( as.numericStats.max, 3 ) + " " +
                                   "1\n");
            } else {
                m_dataWriter.write(Utils.doubleToString( 0, 3 ) + " " +
                                   Utils.doubleToString( as.distinctCount-1, 3 ) + " " +
                                   "1\n");
            }
        }
    }

    private void writeHeaderToMeta(Instances structure) throws IOException
    {
        m_metaWriter.write("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
        m_metaWriter.write("<okcmeta>\n");

        m_metaWriter.write("\t<dimensions value=\"" + structure.numAttributes() +
                           "\" />\n");
        m_metaWriter.write("\t<datapoints value=\"" + structure.numInstances() +
                           "\" />\n");

        for (int i=0; i<structure.numAttributes(); ++i)
        {
            Attribute attr = structure.attribute( i );
            m_metaWriter.write("\t<dimension name=\"" + attr.name() +
                               "\" type=");
            switch (attr.type()) {
            case Attribute.NUMERIC:
                m_metaWriter.write("\"real\" />\n");
                break;
            case Attribute.NOMINAL:
                m_metaWriter.write("\"word\" >\n");
                for (int v=0; v<attr.numValues(); ++v)
                {
                    m_metaWriter.write( "\t\t<value scale=\"" + v + "\">\n" );
                    m_metaWriter.write( "\t\t" + attr.value(v) + "\n" );
                    m_metaWriter.write( "\t\t</value>\n" );
                }
                m_metaWriter.write("\t</dimension>\n");
                break;
            case Attribute.STRING:
                m_metaWriter.write("\"string\" >\n");
                for (int v=0; v<attr.numValues(); ++v)
                {
                    m_metaWriter.write( "\t\t<value scale=\"" + v + "\">\n" );
                    m_metaWriter.write( "\t\t" + attr.value(v) + "\n" );
                    m_metaWriter.write( "\t\t</value>\n" );
                }
                m_metaWriter.write("\t</dimension>\n");
                break;
            default:
                throw new IllegalStateException("Cannot save to this attribute type");
            }
            m_metaWriter.write("\n");
        }

        m_metaWriter.write("</okcmeta>\n");
    }

    /**
     * Save the full data set.
     *
     * @param instances the full data set as an Instances object
     * @throws java.io.IOException if there is an error during saving
     */
    public void saveDataSet(Instances instances) throws Exception
    {
        if ( m_destDataFile == null ) {
            throw new IOException( "No source has been specified" );
        }

        RemoveMissingValues remMissFilter = new RemoveMissingValues();
        remMissFilter.setInputFormat( instances );
        try {
            instances = Filter.useFilter( instances, remMissFilter );
        } catch (Exception e) {
            e.printStackTrace();
            throw e;
        }
        
        if ( !m_isStructureWritten ) {
            saveStructure(instances);
        }

        // save instances
        StringBuffer text = new StringBuffer();
        for (int i=0; i<instances.numInstances(); ++i)
        {
            Instance inst = instances.instance( i );
            for (int j = 0; j < instances.numAttributes(); ++j)
            {
                if (j > 0) text.append(" ");
                text.append(inst.value(j));
            }
            text.append("\n");
        }
        m_dataWriter.write(text.toString());
    }

    public void endSave() throws IOException
    {
        if (m_dataWriter != null)
            m_dataWriter.close();

        if (m_metaWriter != null)
            m_metaWriter.close();
    }
}

⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -