subsetdatastream.java.defunct

Boosting算法软件包

package jboost.tokenizer;

import java.util.*;
import java.io.*;
import java.util.Random;
import jboost.controller.Configuration;
import jboost.tokenizer.*;
import jboost.examples.*;
import jboost.monitor.Monitor;

/** This extension of DataStream generates a pseudo-random subset
    (subsequence) of the stream. Used to seperate data into training
    data and test data.
    @author Yoav Freund
  * @version $Header: /cvsroot/jboost/jboost/src/jboost/tokenizer/SubsetDataStream.java.defunct,v 1.1.1.1 2007/05/16 04:06:02 aarvey Exp $ */
public class SubsetDataStream extends DataStream{
   /** Returns the line number of the current example in the data source.               */
    public long getLineNumber() {return stream.getLineNumber();}

   /** Creates out a string containing the information necessary to 
    *  read the datafile.								*/
    public String toString() {
	return "This is a SubsetDataStream, with seed="+seed
	    +" fraction="+fraction
	    +" selector="+selector
	    +" on top of DataStream:\n"+stream;
    }

   /** Retreives the data specification in a standard internal form.			*/
    public ExampleDescription getExampleDescription()
    {return stream.getExampleDescription();}

   /** Gets an array of attributes in a standard textual form that can
    *  then be transformed into an example by ExampleStream.  */
    public String[] getExampleText() throws ParseException {
	String[] a;
	while(true) {
	    a =stream.getExampleText();
	    if(a==null) return null;
	    int r=Math.abs(generator.nextInt() % 100);
	    if((selector && (r <  fraction)) ||
	       (selector && (r >= fraction))) break;
	} 
	return a;
    }


   /** Using the same example description use this DataStream on a different data 
    *  source.										*/
    public void reset(String data) throws IOException
    {
	stream.reset();
	generator.setSeed(seed);
    }

   /** Constructor 
    * @param stream the underlying stream
    * @param seed a seed number for the random number generator
    * @param selector a boolean selecting which subset we are now
    *                 generating
    * @param fraction an integer specifying the percent (./100) of
    *                  examples we wish to associate with selector =
    *                  "true" */
    public SubsetDataStream(DataStream stream, long seed, 
			    boolean selector, int fraction) 
	throws IOException,SpecFileException
    {
	this.stream = stream;
	this.seed = seed;
	this.selector = selector;
	this.fraction = fraction;

	generator = new Random(seed);
    }
    
//------------------------------- Accessor Methods ------------------------------------//

//------------------------------- Private Members -------------------------------------//

    private long seed;
    private int fraction;
    private boolean selector;
    private DataStream stream;
    private Random generator;

//------------------------------- Test Code -------------------------------------------//\
}