Package org.apache.mahout.vectorizer.encoders

Examples of org.apache.mahout.vectorizer.encoders.Dictionary

49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
    encoder.setTraceDictionary(traceDictionary);
    FeatureVectorEncoder bias = new ConstantValueEncoder("Intercept");
    bias.setTraceDictionary(traceDictionary);
    FeatureVectorEncoder lines = new ConstantValueEncoder("Lines");
    lines.setTraceDictionary(traceDictionary);
    Dictionary newsGroups = new Dictionary();
   
    OnlineLogisticRegression learningAlgorithm =
        new OnlineLogisticRegression(
              20, FEATURES, new L1())
            .alpha(1).stepOffset(1000)
            .decayExponent(0.9)
            .lambda(3.0e-5)
            .learningRate(20);
   
    List<File> files = new ArrayList<File>();
    for (File newsgroup : base.listFiles()) {
      newsGroups.intern(newsgroup.getName());
      files.addAll(Arrays.asList(newsgroup.listFiles()));
    }

    Collections.shuffle(files);
    System.out.printf("%d training files\n", files.size());
View Full Code Here
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
    int leakType = 0;
    if (args.length > 1) {
      leakType = Integer.parseInt(args[1]);
    }

    Dictionary newsGroups = new Dictionary();

    encoder.setProbes(2);
    AdaptiveLogisticRegression learningAlgorithm = new AdaptiveLogisticRegression(20, FEATURES, new L1());
    learningAlgorithm.setInterval(800);
    learningAlgorithm.setAveragingWindow(500);

    List<File> files = Lists.newArrayList();
    File[] directories = base.listFiles();
    Arrays.sort(directories, Ordering.usingToString());
    for (File newsgroup : directories) {
      if (newsgroup.isDirectory()) {
        newsGroups.intern(newsgroup.getName());
        files.addAll(Arrays.asList(newsgroup.listFiles()));
      }
    }
    Collections.shuffle(files);
    System.out.printf("%d training files\n", files.size());
    System.out.printf("%s\n", Arrays.asList(directories));

    double averageLL = 0;
    double averageCorrect = 0;

    int k = 0;
    double step = 0;
    int[] bumps = {1, 2, 5};
    for (File file : files) {
      String ng = file.getParentFile().getName();
      int actual = newsGroups.intern(ng);

      Vector v = encodeFeatureVector(file);
      learningAlgorithm.train(actual, v);

      k++;
View Full Code Here
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
    int leakType = 0;
    if (args.length > 1) {
      leakType = Integer.parseInt(args[1]);
    }

    Dictionary newsGroups = new Dictionary();

    encoder.setProbes(2);
    AdaptiveLogisticRegression learningAlgorithm = new AdaptiveLogisticRegression(20, FEATURES, new L1());
    learningAlgorithm.setInterval(800);
    learningAlgorithm.setAveragingWindow(500);

    List<File> files = Lists.newArrayList();
    for (File newsgroup : base.listFiles()) {
      if (newsgroup.isDirectory()) {
        newsGroups.intern(newsgroup.getName());
        files.addAll(Arrays.asList(newsgroup.listFiles()));
      }
    }
    Collections.shuffle(files);
    System.out.printf("%d training files\n", files.size());

    double averageLL = 0;
    double averageCorrect = 0;

    int k = 0;
    double step = 0;
    int[] bumps = {1, 2, 5};
    for (File file : files.subList(0, 10000)) {
      String ng = file.getParentFile().getName();
      int actual = newsGroups.intern(ng);

      Vector v = encodeFeatureVector(file, actual, leakType);
      learningAlgorithm.train(actual, v);

      k++;
View Full Code Here
108
109
110
111
112
113
114
   * @param typeMap               A map describing the types of the predictor variables.
   */
  public CsvRecordFactory(String targetName, Map<String, String> typeMap) {
    this.targetName = targetName;
    this.typeMap = typeMap;
    targetDictionary = new Dictionary();
  }
View Full Code Here
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
    File base = new File(inputFile);
    //contains the best model
    OnlineLogisticRegression classifier = ModelSerializer.readBinary(new FileInputStream(modelFile), OnlineLogisticRegression.class);


    Dictionary newsGroups = new Dictionary();
    Multiset<String> overallCounts = HashMultiset.create();

    List<File> files = Lists.newArrayList();
    for (File newsgroup : base.listFiles()) {
      if (newsgroup.isDirectory()) {
        newsGroups.intern(newsgroup.getName());
        files.addAll(Arrays.asList(newsgroup.listFiles()));
      }
    }
    System.out.printf("%d test files\n", files.size());
    ResultAnalyzer ra = new ResultAnalyzer(newsGroups.values(), "DEFAULT");
    for (File file : files) {
      String ng = file.getParentFile().getName();

      int actual = newsGroups.intern(ng);
      NewsgroupHelper helper = new NewsgroupHelper();
      Vector input = helper.encodeFeatureVector(file, actual, 0, overallCounts);//no leak type ensures this is a normal vector
      Vector result = classifier.classifyFull(input);
      int cat = result.maxValueIndex();
      double score = result.maxValue();
      double ll = classifier.logLikelihood(actual, input);
      ClassifierResult cr = new ClassifierResult(newsGroups.values().get(cat), score, ll);
      ra.addInstance(newsGroups.values().get(actual), cr);

    }
    output.printf("%s\n\n", ra.toString());
  }
View Full Code Here
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
    int leakType = 0;
    if (args.length > 1) {
      leakType = Integer.parseInt(args[1]);
    }

    Dictionary newsGroups = new Dictionary();

    NewsgroupHelper helper = new NewsgroupHelper();
    helper.getEncoder().setProbes(2);
    AdaptiveLogisticRegression learningAlgorithm = new AdaptiveLogisticRegression(20, NewsgroupHelper.FEATURES, new L1());
    learningAlgorithm.setInterval(800);
    learningAlgorithm.setAveragingWindow(500);

    List<File> files = Lists.newArrayList();
    for (File newsgroup : base.listFiles()) {
      if (newsgroup.isDirectory()) {
        newsGroups.intern(newsgroup.getName());
        files.addAll(Arrays.asList(newsgroup.listFiles()));
      }
    }
    Collections.shuffle(files);
    System.out.printf("%d training files\n", files.size());
    SGDInfo info = new SGDInfo();

    int k = 0;


    for (File file : files) {
      String ng = file.getParentFile().getName();
      int actual = newsGroups.intern(ng);

      Vector v = helper.encodeFeatureVector(file, actual, leakType, overallCounts);
      learningAlgorithm.train(actual, v);

      k++;
View Full Code Here
103
104
105
106
107
108
109
   * @param typeMap               A map describing the types of the predictor variables.
   */
  public CsvRecordFactory(String targetName, Map<String, String> typeMap) {
    this.targetName = targetName;
    this.typeMap = typeMap;
    targetDictionary = new Dictionary();
  }
View Full Code Here
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
    // holds target variable
    List<Integer> target = Lists.newArrayList();

    // for decoding target values
    Dictionary dict = new Dictionary();

    // for permuting data later
    List<Integer> order = Lists.newArrayList();

    for (String line : raw.subList(1, raw.size())) {
      // order gets a list of indexes
      order.add(order.size());

      // parse the predictor variables
      Vector v = new DenseVector(5);
      v.set(0, 1);
      int i = 1;
      Iterable<String> values = onComma.split(line);
      for (String value : Iterables.limit(values, 4)) {
        v.set(i++, Double.parseDouble(value));
      }
      data.add(v);

      // and the target
      target.add(dict.intern(Iterables.get(values, 4)));
    }

    // randomize the order ... original data has each species all together
    // note that this randomization is deterministic
    Random random = RandomUtils.getRandom();
View Full Code Here
108
109
110
111
112
113
114
   * @param typeMap               A map describing the types of the predictor variables.
   */
  public CsvRecordFactory(String targetName, Map<String, String> typeMap) {
    this.targetName = targetName;
    this.typeMap = typeMap;
    targetDictionary = new Dictionary();
  }
View Full Code Here
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
    File base = new File(inputFile);
    //contains the best model
    OnlineLogisticRegression classifier =
        ModelSerializer.readBinary(new FileInputStream(modelFile), OnlineLogisticRegression.class);

    Dictionary newsGroups = new Dictionary();
    Multiset<String> overallCounts = HashMultiset.create();

    List<File> files = Lists.newArrayList();
    for (File newsgroup : base.listFiles()) {
      if (newsgroup.isDirectory()) {
        newsGroups.intern(newsgroup.getName());
        files.addAll(Arrays.asList(newsgroup.listFiles()));
      }
    }
    System.out.println(files.size() + " test files");
    ResultAnalyzer ra = new ResultAnalyzer(newsGroups.values(), "DEFAULT");
    for (File file : files) {
      String ng = file.getParentFile().getName();

      int actual = newsGroups.intern(ng);
      NewsgroupHelper helper = new NewsgroupHelper();
      //no leak type ensures this is a normal vector
      Vector input = helper.encodeFeatureVector(file, actual, 0, overallCounts);
      Vector result = classifier.classifyFull(input);
      int cat = result.maxValueIndex();
      double score = result.maxValue();
      double ll = classifier.logLikelihood(actual, input);
      ClassifierResult cr = new ClassifierResult(newsGroups.values().get(cat), score, ll);
      ra.addInstance(newsGroups.values().get(actual), cr);

    }
    output.println(ra);
  }
View Full Code Here
TOP

Related Classes of org.apache.mahout.vectorizer.encoders.Dictionary

  • mia.classifier.ch14.TrainNewsGroups
  • mia.classifier.ch16.train.TrainNewsGroups
  • org.apache.mahout.classifier.sgd.CsvRecordFactory
  • org.apache.mahout.classifier.sgd.CsvRecordFactoryTest
  • org.apache.mahout.classifier.sgd.OnlineLogisticRegressionTest
  • org.apache.mahout.classifier.sgd.TestASFEmail
  • org.apache.mahout.classifier.sgd.TestNewsGroups
  • org.apache.mahout.classifier.sgd.TrainASFEmail
  • org.apache.mahout.classifier.sgd.TrainNewsGroups
    • Copyright © 2018 www.massapicom. All rights reserved.
    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.