c45.java

Fast implementation of C4/5 in Java

/**
 * @(#)C45.java        1.5.0 09/01/18
 */

package ml.classifier.dt;

import ml.dataset.DataSet;
import ml.dataset.UCIDataSet;
import ml.util.TreeView;
import ml.util.Stopwatch;
import ml.util.HtmlTreeView;
import ml.util.PlainTreeView;

/**
 * The main class of Fast C4.5.
 *
 * @author Ping He
 * @author Xiaohua Xu
 */
public class C45 {
	/**
	 * The c45 decision tree
	 */
	private DecisionTree tree;

	/**
	 * Build a decision tree with the specified data set files.
	 * @param dataSetName the base name of the .names, .data and .test files
	 */
	public C45(String dataSetName){
		DataSet dataSet = new UCIDataSet(dataSetName);
		this.tree = new DecisionTree(dataSet);
	}

	/**
	 * Get the decision tree.
	 */
	public DecisionTree getDecisionTree(){
		return this.tree;
	}

	/**
	 * Build a decision tree with the specified .names and .data files, prune it and print it in
	 * the specified manner. 
	 */
	public static void main(String[] args) {
		C45 c45 = null;
		String dataSetName;
		String output = "plain";
		int repeat = 0;
		TreeView view = null;
		String buildTime = "";
	
		// Illegal input
		if (args.length == 0) {
		    usage();
			return;
		}
		// Ask for help
		if (args.length == 1) {
			if (args[0].equalsIgnoreCase("-h") || args[0].equalsIgnoreCase("-help")){
				usage();
				return;
			}
		}
		
	    dataSetName = args[0];
	    // Interpret the options
		for (int i = 1; i < args.length-1; i += 2) {
			if (args[i].equals("-output") || args[i].equals("-o")) {
				output = args[i+1];
				continue;
			}
			
			if (args[i].equals("-repeat") || args[i].equals("-r")) {
				repeat = Integer.parseInt(args[i+1]);
				continue;
			}
		}

		Stopwatch.start();
		for(int i = 0; i < Math.max(1, repeat); i ++) {
			c45 = new C45(dataSetName);
		}
		Stopwatch.stop();
		
		// Compute the averaged tree building time
		if (repeat > 0) {
		    buildTime = "build time : " + Stopwatch.runtime()/repeat + "ms";
        }
      
		DecisionTree tree = c45.getDecisionTree();
		        
        if (!output.equals("plain") && !output.equals("html")) {
        	System.out.println("Waring:Unsupported Output!");
        	output = "plain";
        }
        
        if(output.equals("plain")) {
        	view = plainView(tree, buildTime);
        }
        
        if(output.equals("html")) {
        	view = htmlView(tree, buildTime);
        }
        
		System.out.println(view);
 	}

	private static TreeView plainView(DecisionTree tree, String buildTime){
		TreeView v1 = new PlainTreeView(tree);
		v1.insert("Tree Before Pruning");
		v1.append("tree size : " + tree.size());
		v1.append("train error : " + tree.getTrainError());
		v1.append(buildTime+TreeView.CR);
		
		tree.prune();
		
		TreeView v2 = new PlainTreeView(tree);
		v2.insert("Tree After Pruning");
		v2.append("tree size : " + tree.size());
		v2.append("train error : " + tree.getTrainError());
		
		return v1.union(v2);
	}
 
	private static TreeView htmlView(DecisionTree tree, String buildTime) {
		TreeView v1 = new HtmlTreeView(tree);
		v1.insert("<h2>Tree Before Pruning</h2>");
		v1.append("<p>tree size : "+tree.size()+"<br>");
		v1.append("train error : "+tree.getTrainError()+"<br>");
		v1.append(buildTime+"</p>").append(TreeView.CR);
		
		tree.prune();
		
		TreeView v2 = new HtmlTreeView(tree);
		v2.insert("<h2>Tree After Pruning</h2>");
		v2.append("<p>tree size : "+tree.size()+"<br>");
		v2.append("train error : "+tree.getTrainError()+"</p>");
		
		return v1.union(v2);
	}
		
 	private static void usage(String... messages) {
 	    System.out.println("Usage: java " + C45.class + " dataSetName [-output plain | html -repeat times]");
 	    for (String line : messages) {
 	    	System.out.println(line);
 	    }
	}
}