Package org.apache.mahout.classifier.df.data

Examples of org.apache.mahout.classifier.df.data.Data

143
144
145
146
147
148
149
150
151
152
153
  @Override
  protected void cleanup(Context context) throws IOException, InterruptedException {
    // prepare the data
    log.debug("partition: {} numInstances: {}", partition, instances.size());
   
    Data data = new Data(getDataset(), instances);
    Bagging bagging = new Bagging(getTreeBuilder(), data);
   
    TreeID key = new TreeID();
   
    log.debug("Building {} trees", nbTrees);
View Full Code Here
143
144
145
146
147
148
149
150
151
152
153
  @Override
  protected void cleanup(Context context) throws IOException, InterruptedException {
    // prepare the data
    log.debug("partition: {} numInstances: {}", partition, instances.size());
   
    Data data = new Data(getDataset(), instances);
    Bagging bagging = new Bagging(getTreeBuilder(), data);
   
    TreeID key = new TreeID();
   
    log.debug("Building {} trees", nbTrees);
View Full Code Here
64
65
66
67
68
69
70
71
72
73
    super.setup(context);
   
    Configuration conf = context.getConfiguration();
   
    log.info("Loading the data...");
    Data data = loadData(conf, getDataset());
    log.info("Data loaded : {} instances", data.size());
   
    bagging = new Bagging(getTreeBuilder(), data);
  }
View Full Code Here
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
    Node childNode;
    if (data.getDataset().isNumerical(best.getAttr())) {
      boolean[] temp = null;

      Data loSubset = data.subset(Condition.lesser(best.getAttr(), best.getSplit()));
      Data hiSubset = data.subset(Condition.greaterOrEquals(best.getAttr(), best.getSplit()));

      if (loSubset.isEmpty() || hiSubset.isEmpty()) {
        // the selected attribute did not change the data, avoid using it in the child notes
        selected[best.getAttr()] = true;
      } else {
        // the data changed, so we can unselect all previousely selected NUMERICAL attributes
        temp = selected;
        selected = cloneCategoricalAttributes(data.getDataset(), selected);
      }

      // size of the subset is less than the minSpitNum
      if (loSubset.size() < minSplitNum || hiSubset.size() < minSplitNum) {
        // branch is not split
        double label;
        if (data.getDataset().isNumerical(data.getDataset().getLabelId())) {
          label = sum / data.size();
        } else {
View Full Code Here
83
84
85
86
87
88
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
   *          number of trees to grow
   */
  private void runIteration(Random rng, Data data, int m, int nbtrees) {
   
    log.info("Splitting the data");
    Data train = data.clone();
    Data test = train.rsplit(rng, (int) (data.size() * 0.1));
   
    DefaultTreeBuilder treeBuilder = new DefaultTreeBuilder();
   
    SequentialBuilder forestBuilder = new SequentialBuilder(rng, treeBuilder, train);
   
    // grow a forest with m = log2(M)+1
    treeBuilder.setM(m);
   
    long time = System.currentTimeMillis();
    log.info("Growing a forest with m={}", m);
    DecisionForest forestM = forestBuilder.build(nbtrees);
    sumTimeM += System.currentTimeMillis() - time;
    numNodesM += forestM.nbNodes();
   
    // grow a forest with m=1
    treeBuilder.setM(1);
   
    time = System.currentTimeMillis();
    log.info("Growing a forest with m=1");
    DecisionForest forestOne = forestBuilder.build(nbtrees);
    sumTimeOne += System.currentTimeMillis() - time;
    numNodesOne += forestOne.nbNodes();
   
    // compute the test set error (Selection Error), and mean tree error (One Tree Error),
    double[] testLabels = test.extractLabels();
    double[] predictions = new double[test.size()];
   
    forestM.classify(test, predictions);
    sumTestErrM += ErrorEstimate.errorRate(testLabels, predictions);
   
    forestOne.classify(test, predictions);
View Full Code Here
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
    }
   
    // load the data
    FileSystem fs = dataPath.getFileSystem(new Configuration());
    Dataset dataset = Dataset.load(getConf(), datasetPath);
    Data data = DataLoader.loadData(dataset, fs, dataPath);
   
    // take m to be the first integer less than log2(M) + 1, where M is the
    // number of inputs
    int m = (int) Math.floor(Maths.log(2, data.getDataset().nbAttributes()) + 1);
   
    Random rng = RandomUtils.getRandom();
    for (int iteration = 0; iteration < nbIterations; iteration++) {
      log.info("Iteration {}", iteration);
      runIteration(rng, data, m, nbTrees);
View Full Code Here
241
242
243
244
245
246
247
248
249
250
  }
 
  protected static Data loadData(Configuration conf, Path dataPath, Dataset dataset) throws IOException {
    log.info("Loading the data...");
    FileSystem fs = dataPath.getFileSystem(conf);
    Data data = DataLoader.loadData(dataset, fs, dataPath);
    log.info("Data Loaded");
   
    return data;
  }
View Full Code Here
243
244
245
246
247
248
249
250
251
252
  }
 
  protected static Data loadData(Configuration conf, Path dataPath, Dataset dataset) throws IOException {
    log.info("Loading the data...");
    FileSystem fs = dataPath.getFileSystem(conf);
    Data data = DataLoader.loadData(dataset, fs, dataPath);
    log.info("Data Loaded");
   
    return data;
  }
View Full Code Here
140
141
142
143
144
145
146
147
148
149
150
  @Override
  protected void cleanup(Context context) throws IOException, InterruptedException {
    // prepare the data
    log.debug("partition: {} numInstances: {}", partition, instances.size());
   
    Data data = new Data(getDataset(), instances);
    Bagging bagging = new Bagging(getTreeBuilder(), data);
   
    TreeID key = new TreeID();
   
    log.debug("Building {} trees", nbTrees);
View Full Code Here
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
  private static Data[] generateTrainingDataA() throws DescriptorException {
    // Dataset
    Dataset dataset = DataLoader.generateDataset("C N N C L", false, TRAIN_DATA);
   
    // Training data
    Data data = DataLoader.loadData(dataset, TRAIN_DATA);
    @SuppressWarnings("unchecked")
    List<Instance>[] instances = new List[3];
    for (int i = 0; i < instances.length; i++) {
      instances[i] = Lists.newArrayList();
    }
    for (int i = 0; i < data.size(); i++) {
      if (data.get(i).get(0) == 0.0d) {
        instances[0].add(data.get(i));
      } else {
        instances[1].add(data.get(i));
      }
    }
    Data[] datas = new Data[instances.length];
    for (int i = 0; i < datas.length; i++) {
      datas[i] = new Data(dataset, instances[i]);
    }

    return datas;
  }
View Full Code Here
TOP

Related Classes of org.apache.mahout.classifier.df.data.Data

  • org.apache.mahout.classifier.df.Bagging
  • org.apache.mahout.classifier.df.BreimanExample
  • org.apache.mahout.classifier.df.builder.DecisionTreeBuilder
  • org.apache.mahout.classifier.df.builder.DefaultTreeBuilder
  • org.apache.mahout.classifier.df.builder.InfiniteRecursionTest
  • org.apache.mahout.classifier.df.DecisionForestTest
  • org.apache.mahout.classifier.df.mapreduce.BuildForest
  • org.apache.mahout.classifier.df.mapreduce.inmem.InMemMapper
  • org.apache.mahout.classifier.df.mapreduce.partial.Step1Mapper
  • org.apache.mahout.classifier.df.split.DefaultIgSplit
    • 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.