sampling.java

数据挖掘中的Aprior算法java实现

package datamining;

import java.io.*;
import java.util.*;

/**
 * Class for finding frequent itemsets using sampling
 * with the Apriori algorithm. 
 *
 * @author Michael Holler
 * @version 0.1, 16.03.2004
 */
public class Sampling {

  String filename;	// name of the database file
  int minsup;		// the minimal support threshold
  int samplesize;	// how big a part is taken as a sample
  Item root;
  DataHandler dh;
  
  /**
   * Default constructur for creating a Sampling object.
   */ 
  public Sampling() {
  }
  
  /**
   * Constructor for creating a Sampling object with parameters.
   *
   * @param	filename	the name of database file
   * @param	support		the minimal support threshold
   * @param	samplesize	how manieth row will be taken to the example
   */
  public Sampling(String filename, int support, int samplesize) { 
    this.dh = new DataHandler(filename); 
    this.minsup = support;
    this.samplesize = samplesize;
  }

  /**
   * The workhorse method which includes the basic idea of the
   * sampling algorithm. 
   */
  public void findFrequentSets() {
    // mark a sample and run it with apriori
    dh.setSample(samplesize);
    Apriori apriori = new Apriori(minsup/samplesize, dh); 
    apriori.findFrequentSets();

    // get the results and count the true support over the whole database
    this.root = apriori.getTrie();
    int candidates = Tools.zeroSupport(this.root);
    dh.setSample(1);
    Tools.countSupport(this.root, dh);

    // remove infrequent
    int pruned = Tools.pruneCandidates(root, this.minsup);
    int found = Tools.zeroSupport(root);
    System.out.println("candidates: " + candidates +
		       ", pruned: " + pruned +
		       ", found: " + found);
  }

  /**
   * Main method for testing the algorithm.
   *
   * @param 	args	the arguments can contain the filename
   * 			of the testfile and the minimal support
   * 			threshold and a filename for output
   */
  public static void main(String args[]) {
    String testfile = "test.dat";
    String outfile = "";
    int support = 5;
    int samplesize = 10;
    try {
      testfile = args[0];
    } catch (Exception e) {
      System.out.println("Didn't get filename. Using '" + testfile + "'.");
    }
    try {
      support = new Integer(args[1]).intValue();
    } catch (Exception e) {
      System.out.println("Didn't get support threshold. Using '" + support + "'.");
    }	    
    try {
      samplesize = new Integer(args[2]).intValue();
    } catch (Exception e) {
      System.out.println("Didn't get output filename. Not printing.");
    }

    StopWatch sw = new StopWatch();
    sw.start();
    Sampling aws = new Sampling(testfile, support, samplesize);
    aws.findFrequentSets();
    sw.stop();
    sw.print();
  }
}