机器学习

import java.util.ArrayList;

import cc.mallet.types.*;
import cc.mallet.classify.Classifier;
import cc.mallet.classify.ClassifierTrainer;
import cc.mallet.classify.NaiveBayesTrainer;
import cc.mallet.classify.Trial;
import cc.mallet.pipe.*;
import cc.mallet.pipe.iterator.*;

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

public class SpamDetector {

	public static void main（String[] args）{
		
    	String stopListFilePath = “data/stoplists/en.txt“;
    	String dataFolderPath = “data/ex6DataEmails/train“;
    	String testFolderPath = “data/ex6DataEmails/test“;
    	
		ArrayList pipeList = new ArrayList（）;
		pipeList.add（new Input2CharSequence（“UTF-8“））;
		Pattern tokenPattern = Pattern.compile（“[\p{L}\p{N}_]+“）;
		pipeList.add（new CharSequence2TokenSequence（tokenPattern））;
		pipeList.add（new TokenSequenceLowercase（））;
		pipeList.add（new TokenSequenceRemoveStopwords（new File（stopListFilePath） “utf-8“ false false false））;
		pipeList.add（new TokenSequence2FeatureSequence（））;
		pipeList.add（new FeatureSequence2FeatureVector（））;
		pipeList.add（new Target2Label（））;
		SerialPipes pipeline = new SerialPipes（pipeList）;
		
		FileIterator folderIterator = new FileIterator（
				new File[] {new File（dataFolderPath）}
		         new TxtFilter（）
		         FileIterator.LAST_DIRECTORY）;

		
		InstanceList instances = new InstanceList（pipeline）;
		
		instances.addThruPipe（folderIterator）;
		
		ClassifierTrainer classifierTrainer = new NaiveBayesTrainer（）;
		Classifier classifier = classifierTrainer.train（instances）;

		InstanceList testInstances = new InstanceList（classifier.getInstancePipe（））;
		folderIterator = new FileIterator（
				new File[] {new File（testFolderPath）}
		         new TxtFilter（）
		         FileIterator.LAST_DIRECTORY）;
        testInstances.addThruPipe（folderIterator）;
        
        Trial trial = new Trial（classifier testInstances）;
        
        System.out.println（“Accuracy: “ + trial.getAccuracy（））;
        System.out.println（“F1 for class ‘spam‘: “ + trial.getF1（“spam“））;

        System.out.println（“Precision for class ‘“ +
                           classifier.getLabelAlphabet（）.lookupLabel（1） + “‘: “ +
                           trial.getPrecision（1））;

        System.out.println（“Recall for class ‘“ +
                           classifier.getLabelAlphabet（）.lookupLabel（1） + “‘: “ +
                           trial.getRecall（1））;

		
		

	}
}

 属性            大小     日期    时间   名称
----------- ---------  ---------- -----  ----
     文件      126894  2016-04-26 13:55  MachineLearninginJava_CodeB03686_Software list.pdf
     目录           0  2016-04-26 14:00  MachineLearninginJava_CodeCode
     文件         417  2016-04-03 07:21  MachineLearninginJava_CodeCode.gitignore
     目录           0  2016-04-26 13:44  MachineLearninginJava_CodeCodeMLJ-Chapter10
     文件         402  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10.classpath
     文件         372  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10.project
     目录           0  2016-04-26 13:44  MachineLearninginJava_CodeCodeMLJ-Chapter10data
     目录           0  2016-04-26 13:44  MachineLearninginJava_CodeCodeMLJ-Chapter10databc
     目录           0  2016-04-26 13:43  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness
     文件        2560  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness01.txt
     文件        2252  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness02.txt
     文件        1552  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness03.txt
     文件        2412  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness04.txt
     文件        1570  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness05.txt
     文件        1187  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness06.txt
     文件        1669  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness07.txt
     文件        1922  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness08.txt
     文件        1494  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness09.txt
     文件        1449  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness10.txt
     文件        1144  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness11.txt
     文件        1847  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness12.txt
     文件        1830  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness13.txt
     文件        2981  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness14.txt
     文件        3808  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness15.txt
     文件        1393  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness16.txt
     文件        1299  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness17.txt
     文件        1002  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness18.txt
     文件        1733  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness19.txt
     文件        3854  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness20.txt
     文件        2046  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness21.txt
     文件        1933  2016-04-03 07:21  MachineLearninginJava_CodeCodeMLJ-Chapter10databcusiness22.txt
............此处省略3343个文件信息