Examples of FileLineIterable

net.myrrix.common.iterator.FileLineIterable
Iterable representing the lines of a text file. It can produce an {@link Iterator} over those lines. Thisassumes the text file's lines are delimited in a manner consistent with how {@link java.io.BufferedReader}defines lines. This class will uncompress files that end in .zip or .gz accordingly, too. @author Sean Owen @author Mahout @since 1.0
org.apache.mahout.cf.taste.impl.common.FileLineIterable
Iterable representing the lines of a text file. It can produce an {@link Iterator} over those lines.This assumes the text file's lines are delimited in a manner consistent with how {@link java.io.BufferedReader} defines lines.This class will uncompress files that end in .zip or .gz accordingly, too.
org.apache.mahout.common.FileLineIterable
Iterable representing the lines of a text file. It can produce an {@link Iterator} over those lines. Thisassumes the text file's lines are delimited in a manner consistent with how {@link java.io.BufferedReader}defines lines. This class will uncompress files that end in .zip or .gz accordingly, too.
org.apache.mahout.common.iterator.FileLineIterable
Iterable representing the lines of a text file. It can produce an {@link Iterator} over those lines. Thisassumes the text file's lines are delimited in a manner consistent with how {@link java.io.BufferedReader}defines lines.

Examples of org.apache.mahout.common.iterator.FileLineIterable

      FPGrowth<String> fp1 = new org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth<String>();


    final Map<Set<String>,Long> results1 = Maps.newHashMap();
    
    fp1.generateTopKFrequentPatterns(
      new StringRecordIterator(new FileLineIterable(Resources.getResource(inputFilename).openStream()), "\\s+"),


      fp1.generateFList(new StringRecordIterator(new FileLineIterable(Resources.getResource(inputFilename)
           .openStream()), "\\s+"), minSupport), minSupport, 100000, 
      returnableFeatures,
      new OutputCollector<String,List<Pair<List<String>,Long>>>() {
        
        @Override
        public void collect(String key, List<Pair<List<String>,Long>> value) {
          
          for (Pair<List<String>,Long> v : value) {
            List<String> l = v.getFirst();
            results1.put(new HashSet<String>(l), v.getSecond());
            log.info("found pat ["+v.getSecond()+"]: "+ v.getFirst());
          }
        }
        
      }, new StatusUpdater() {
        
        @Override
        public void update(String status) {}
      });


    FPGrowthObj<String> fp2 = new FPGrowthObj<String>();
    final Map<Set<String>,Long> initialResults2 = Maps.newHashMap();
    fp2.generateTopKFrequentPatterns(
      new StringRecordIterator(new FileLineIterable(Resources.getResource(inputFilename).openStream()), "\\s+"),


      fp2.generateFList(new StringRecordIterator(new FileLineIterable(Resources.getResource(inputFilename)
           .openStream()), "\\s+"), minSupport), minSupport, 100000, 
      new HashSet<String>(),
      new OutputCollector<String,List<Pair<List<String>,Long>>>() {
        
        @Override

View Full Code Here

Examples of org.apache.mahout.common.iterator.FileLineIterable


  @Test
  public void testSpecificCaseFromRetailDataMinSup500() throws IOException {
    FPGrowthObj<String> fp = new FPGrowthObj<String>();
    
    StringRecordIterator it = new StringRecordIterator(new FileLineIterable(Resources.getResource(
      "retail.dat").openStream()), "\\s+");
    int pattern_41_36_39 = 0;
    while (it.hasNext()) {
      Pair<List<String>,Long> next = it.next();
      List<String> items = next.getFirst();
      if (items.contains("41") && items.contains("36") && items.contains("39")) {
        pattern_41_36_39++;
      }
    }
    
    final Map<Set<String>,Long> results = Maps.newHashMap();
    
    Set<String> returnableFeatures = new HashSet<String>();
    returnableFeatures.add("41");
    returnableFeatures.add("36");
    returnableFeatures.add("39");
    
    fp.generateTopKFrequentPatterns(
      new StringRecordIterator(new FileLineIterable(Resources.getResource("retail.dat").openStream()), "\\s+"),


      fp.generateFList(new StringRecordIterator(new FileLineIterable(Resources.getResource("retail.dat")
          .openStream()), "\\s+"), 500), 500, 1000, returnableFeatures,
      new OutputCollector<String,List<Pair<List<String>,Long>>>() {
        
        @Override
        public void collect(String key, List<Pair<List<String>,Long>> value) {

View Full Code Here

Examples of org.apache.mahout.common.iterator.FileLineIterable

    File input = new File(inputDir, "test.txt");
    params.set(PFPGrowth.INPUT, input.getAbsolutePath());
    params.set(PFPGrowth.OUTPUT, outputDir.getAbsolutePath());
    Writer writer = Files.newWriter(input, Charsets.UTF_8);
    try {
      StringRecordIterator it = new StringRecordIterator(new FileLineIterable(Resources.getResource(
        "retail.dat").openStream()), "\\s+");
      Collection<List<String>> transactions = Lists.newArrayList();
      
      while (it.hasNext()) {
        Pair<List<String>,Long> next = it.next();

View Full Code Here

Examples of org.apache.mahout.common.iterator.FileLineIterable

  /**
   * Test Parallel FPGrowth on retail data using top-level runPFPGrowth() method
   */ 
  @Test
  public void testRetailDataMinSup100() throws Exception {
    StringRecordIterator it = new StringRecordIterator(new FileLineIterable(Resources.getResource(
      "retail_results_with_min_sup_100.dat").openStream()), "\\s+");
    Map<Set<String>,Long> expectedResults = Maps.newHashMap();
    while (it.hasNext()) {
      Pair<List<String>,Long> next = it.next();
      List<String> items = Lists.newArrayList(next.getFirst());

View Full Code Here

Examples of org.apache.mahout.common.iterator.FileLineIterable

  /**
   * Test Parallel FPG on retail data, running various stages individually
   */ 
  @Test
  public void testRetailDataMinSup100InSteps() throws Exception {
    StringRecordIterator it = new StringRecordIterator(new FileLineIterable(Resources.getResource(
      "retail_results_with_min_sup_100.dat").openStream()), "\\s+");    
    Map<Set<String>,Long> expectedResults = Maps.newHashMap();
    while (it.hasNext()) {
      Pair<List<String>,Long> next = it.next();
      List<String> items = Lists.newArrayList(next.getFirst());

View Full Code Here

Examples of org.apache.mahout.common.iterator.FileLineIterable


    MathHelper.assertMatrixEquals(expectedAdjacencyMatrix, actualAdjacencyMatrix);
    MathHelper.assertMatrixEquals(expectedTransitionMatrix, actualTransitionMatrix);


    Map<Long,Double> steadyStateProbabilities = Maps.newHashMap();
    for (CharSequence line : new FileLineIterable(new File(outputDir, "part-m-00000"))) {
      String[] tokens = Iterables.toArray(Splitter.on("\t").split(line), String.class);
      steadyStateProbabilities.put(Long.parseLong(tokens[0]), Double.parseDouble(tokens[1]));
    }


    assertEquals(4, steadyStateProbabilities.size());

View Full Code Here

Examples of org.apache.mahout.common.iterator.FileLineIterable

    File input = new File(inputDir, "synth_test.txt");
    params.set(PFPGrowth.INPUT, input.getAbsolutePath());
    params.set(PFPGrowth.OUTPUT, outputDir.getAbsolutePath());
    Writer writer = Files.newWriter(input, Charsets.UTF_8);
    try {
      StringRecordIterator it = new StringRecordIterator(new FileLineIterable(Resources.getResource(
        "FPGsynth.dat").openStream()), "\\s+");
      Collection<List<String>> transactions = Lists.newArrayList();
      
      while (it.hasNext()) {
        Pair<List<String>,Long> next = it.next();

View Full Code Here

Examples of org.apache.mahout.common.iterator.FileLineIterable


    final Map<Set<String>,Long> seqResult = Maps.newHashMap();
    
    FPGrowthObj<String> fpSeq = new FPGrowthObj<String>();
    fpSeq.generateTopKFrequentPatterns(
      new StringRecordIterator(new FileLineIterable(Resources.getResource(inputFilename).openStream()), "\\s+"),


      fpSeq.generateFList(new StringRecordIterator(new FileLineIterable(Resources.getResource(inputFilename)
           .openStream()), "\\s+"), minSupport), minSupport, 1000000, 
      null,
      new OutputCollector<String,List<Pair<List<String>,Long>>>() {
        
        @Override

View Full Code Here

Examples of org.apache.mahout.common.iterator.FileLineIterable


    MathHelper.assertMatrixEquals(expectedAdjacencyMatrix, actualAdjacencyMatrix);
    MathHelper.assertMatrixEquals(expectedTransitionMatrix, actualTransitionMatrix);


    Map<Integer,Double> rankPerVertex = Maps.newHashMapWithExpectedSize(numVertices);
    for (CharSequence line : new FileLineIterable(new File(outputDir, "part-m-00000"))) {
      String[] tokens = Iterables.toArray(Splitter.on("\t").split(line), String.class);
      rankPerVertex.put(Integer.parseInt(tokens[0]), Double.parseDouble(tokens[1]));
    }


    assertEquals(4, rankPerVertex.size());

View Full Code Here

Examples of org.apache.mahout.common.iterator.FileLineIterable

    FPGrowthObj<String> fp = new FPGrowthObj<String>();
    
    String inputFilename = "FPGsynth.dat";
    int minSupport = 50;


    StringRecordIterator it = new StringRecordIterator(new FileLineIterable(Resources.getResource(
                                                                                                  inputFilename).openStream()), "\\s+");
    int patternCnt_10_13_1669 = 0;
    int patternCnt_10_13 = 0;
    while (it.hasNext()) {
      Pair<List<String>,Long> next = it.next();
      List<String> items = next.getFirst();
      if (items.contains("10") && items.contains("13")) {
        patternCnt_10_13++;
        if (items.contains("1669")) {
          patternCnt_10_13_1669++;
        }
      }
    }
    
    if (patternCnt_10_13_1669 < minSupport) 
      throw new IllegalStateException("the test is broken or data is missing ("
                                      + patternCnt_10_13_1669+", "
                                      + patternCnt_10_13+")");


    final Map<Set<String>,Long> results = Maps.newHashMap();
    
    Set<String> features_10_13 = new HashSet<String>();
    features_10_13.add("10");
    features_10_13.add("13");


    Set<String> returnableFeatures = new HashSet<String>();
    returnableFeatures.add("10");
    returnableFeatures.add("13");
    returnableFeatures.add("1669");
    
    fp.generateTopKFrequentPatterns(
                                    new StringRecordIterator(new FileLineIterable(Resources.getResource(inputFilename).openStream()), "\\s+"),


                                    fp.generateFList(new StringRecordIterator(new FileLineIterable(Resources.getResource(inputFilename)
                                                                                                   .openStream()), "\\s+"), minSupport), minSupport, 100000, 
                                    returnableFeatures,
                                    new OutputCollector<String,List<Pair<List<String>,Long>>>() {
        
                                      @Override

View Full Code Here

0 1 2 3 4 5

TOP

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.