Package org.apache.mahout.df.data

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

147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
  @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);
    SingleTreePredictions callback = null;
    int[] predictions = null;
    for (int treeId = 0; treeId < nbTrees; treeId++) {
      log.debug("Building tree number : {}", treeId);
      if (isOobEstimate() && !isNoOutput()) {
        callback = new SingleTreePredictions(data.size());
        predictions = callback.getPredictions();
      }
     
      Node tree = bagging.build(treeId, rng, callback);
     
View Full Code Here
58
59
60
61
62
63
64
65
66
67
68
   *           if the data is not set
   */
  public Node build(int treeId, Random rng, PredictionCallback callback) {
    log.debug("Bagging...");
    Arrays.fill(sampled, false);
    Data bag = data.bagging(rng, sampled);
   
    log.debug("Building...");
    Node tree = treeBuilder.build(rng, bag);
   
    // predict the label for the out-of-bag elements
View Full Code Here
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
      log.warn("attribute {} already selected in a parent node", best.attr);
    }
   
    Node childNode;
    if (data.getDataset().isNumerical(best.attr)) {
      Data loSubset = data.subset(Condition.lesser(best.attr, best.split));
      Node loChild = build(rng, loSubset);
     
      Data hiSubset = data.subset(Condition.greaterOrEquals(best.attr, best.split));
      Node hiChild = build(rng, hiSubset);
     
      childNode = new NumericalNode(best.attr, best.split, loChild, hiChild);
    } else { // CATEGORICAL attribute
      selected[best.attr] = true;
     
      double[] values = data.values(best.attr);
      Node[] childs = new Node[values.length];
     
      for (int index = 0; index < values.length; index++) {
        Data subset = data.subset(Condition.equals(best.attr, values[index]));
        childs[index] = build(rng, subset);
      }
     
      childNode = new CategoricalNode(best.attr, values, childs);
     
View Full Code Here
65
66
67
68
69
70
71
72
73
74
75
    double hy = entropy(data); // H(Y)
    double hyx = 0.0; // H(Y|X)
    double invDataSize = 1.0 / data.size();
   
    for (double value : values) {
      Data subset = data.subset(Condition.equals(attr, value));
      hyx += subset.size() * invDataSize * entropy(subset);
    }
   
    return hy - hyx;
  }
View Full Code Here
85
86
87
88
89
90
91
92
93
94
95
96
97
98
  protected double numericalIg(Data data, int attr, double split) {
    double hy = entropy(data);
    double invDataSize = 1.0 / data.size();
   
    // LO subset
    Data subset = data.subset(Condition.lesser(attr, split));
    hy -= subset.size() * invDataSize * entropy(subset);
   
    // HI subset
    subset = data.subset(Condition.greaterOrEquals(attr, split));
    hy -= subset.size() * invDataSize * entropy(subset);
   
    return hy;
  }
View Full Code Here
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
  @Override
  public void close() throws IOException {
    // 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);
    SingleTreePredictions callback = null;
    int[] predictions = null;
    for (int treeId = 0; treeId < nbTrees; treeId++) {
      log.debug("Building tree number: {}", treeId);
      if (isOobEstimate() && !isNoOutput()) {
        callback = new SingleTreePredictions(data.size());
        predictions = callback.getPredictions();
      }
     
      Node tree = bagging.build(treeId, rng, callback);
     
View Full Code Here
37
38
39
40
41
42
43
44
45
46
47
    // all the vectors have the same label (0)
    double[][] temp = Utils.randomDoublesWithSameLabel(rng, descriptor, 100, 0);
    String[] sData = Utils.double2String(temp);
    Dataset dataset = DataLoader.generateDataset(descriptor, sData);
    Data data = DataLoader.loadData(dataset, sData);
    DefaultIgSplit iG = new DefaultIgSplit();

    double expected = 0.0 - 1.0 * Math.log(1.0) / Math.log(2.0);
    assertEquals(expected, iG.entropy(data));

View Full Code Here
35
36
37
38
39
40
41
42
43
44
45
46
47
48
    IgSplit ref = new DefaultIgSplit();
    IgSplit opt = new OptIgSplit();

    Random rng = RandomUtils.getRandom();
    Data data = Utils.randomData(rng, nbAttributes, numInstances);

    for (int nloop = 0; nloop < n; nloop++) {
      int attr = rng.nextInt(data.getDataset().nbAttributes());
      // System.out.println("IsNumerical: " + data.dataset.isNumerical(attr));

      Split expected = ref.computeSplit(data, attr);
      Split actual = opt.computeSplit(data, attr);
View Full Code Here
67
68
69
70
71
72
73
74
75
76
77
    // store the data into a file
    String[] sData = Utils.double2String(source);
    Path dataPath = Utils.writeDataToTestFile(sData);
    Dataset dataset = DataLoader.generateDataset(descriptor, sData);
    Data data = DataLoader.loadData(dataset, sData);

    JobConf jobConf = new JobConf();
    jobConf.setNumMapTasks(numMaps);

    // prepare a custom TreeBuilder that will classify each
View Full Code Here
199
200
201
202
203
204
205
206
207
208
  }
 
  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
TOP

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

  • org.apache.hadoop.fs.FSDataInputStream
  • org.apache.mahout.common.FileLineIterator
  • org.apache.mahout.common.iterator.FileLineIterator
  • org.apache.mahout.df.Bagging
  • org.apache.mahout.df.BreimanExample
  • org.apache.mahout.df.builder.DefaultTreeBuilder
  • org.apache.mahout.df.builder.DefaultTreeBuilderTest
  • org.apache.mahout.df.builder.InfiniteRecursionTest
  • org.apache.mahout.df.mapred.BuildForest
  • org.apache.mahout.df.mapred.partial.PartitionBugTest
    • 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.