Package org.apache.mahout.df.node

Examples of org.apache.mahout.df.node.Node

79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
        best = split;
    }

    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);

      return new NumericalNode(best.attr, best.split, loChild, hiChild);
    } else { // CATEGORICAL attribute
      double[] values = data.values(best.attr);
      Node[] childs = new Node[values.length];
View Full Code Here
99
100
101
102
103
104
105
106
107
108
109
    }

    initRandom((InMemInputSplit)context.getInputSplit());

    log.debug("Building...");
    Node tree = bagging.build(key.get(), rng, callback);

    if (!isNoOutput()) {
      log.debug("Outputing...");
      MapredOutput mrOut = new MapredOutput(tree, predictions);
View Full Code Here
108
109
110
111
112
113
114
115
116
117
118
    }

    initRandom(split);

    log.debug("Building...");
    Node tree = bagging.build(key.get(), rng, callback);

    if (!isNoOutput()) {
      log.debug("Outputing...");
      MapredOutput mrOut = new MapredOutput(tree, predictions);
View Full Code Here
161
162
163
164
165
166
167
168
169
170
171
      if (isOobEstimate() && !isNoOutput()) {
        callback = new SingleTreePredictions(data.size());
        predictions = callback.getPredictions();
      }

      Node tree = bagging.build(treeId, rng, callback);

      key.set(partition, firstTreeId + treeId);

      if (!isNoOutput()) {
        MapredOutput emOut = new MapredOutput(tree, predictions);
View Full Code Here
134
135
136
137
138
139
140
141
142
143
144
      partitions.add(partition);

      int nbTrees = Step1Mapper.nbTrees(numMaps, numTrees, partition);

      for (int treeId = 0; treeId < nbTrees; treeId++) {
        Node tree = new Leaf(rng.nextInt(100));

        keys[index] = new TreeID(partition, treeId);
        values[index] = new MapredOutput(tree, nextIntArray(rng, numInstances));

        index++;
View Full Code Here
134
135
136
137
138
139
140
141
142
143
144
      partitions.add(partition);

      int nbTrees = Step1Mapper.nbTrees(numMaps, numTrees, partition);

      for (int treeId = 0; treeId < nbTrees; treeId++) {
        Node tree = new Leaf(rng.nextInt(100));

        keys[index] = new TreeID(partition, treeId);
        values[index] = new MapredOutput(tree, nextIntArray(rng, numInstances));

        index++;
View Full Code Here
143
144
145
146
147
148
149
150
151
152
153
      // load (key, tree)
      TreeID key = new TreeID();
      for (int index = 0; index < nbTrees; index++) {
        key.readFields(in);
        Node value = Node.read(in);

        assertEquals("index: " + index, keys[index], key);
        assertEquals("index: " + index, trees[index], value);
      }
    } finally {
View Full Code Here
67
68
69
70
71
72
73
74
75
76
77
78
79
    if (data.isEmpty()) {
      return; // nothing to classify
    }

    for (int treeId = 0; treeId < trees.size(); treeId++) {
      Node tree = trees.get(treeId);

      for (int index = 0; index < data.size(); index++) {
        int prediction = tree.classify(data.get(index));
        callback.prediction(treeId, index, prediction);
      }
    }
  }
View Full Code Here
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
    if (alreadySelected) {
      // attribute already selected
      log.warn("attribute {} already selected in a parent node", best.getAttr());
    }
   
    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);
      }

      Node loChild = build(rng, loSubset);
      Node hiChild = build(rng, hiSubset);

      // restore the selection state of the attributes
      if (temp != null) {
        selected = temp;
      } else {
View Full Code Here
TOP

Related Classes of org.apache.mahout.df.node.Node

  • org.apache.mahout.df.Bagging
  • org.apache.mahout.df.builder.DefaultTreeBuilder
  • org.apache.mahout.df.DecisionForest
  • org.apache.mahout.df.mapred.inmem.InMemMapper
  • org.apache.mahout.df.mapred.partial.PartialBuilderTest
  • org.apache.mahout.df.mapred.partial.Step1Mapper
  • org.apache.mahout.df.mapreduce.inmem.InMemMapper
  • org.apache.mahout.df.mapreduce.partial.InterResultsTest
  • org.apache.mahout.df.mapreduce.partial.PartialBuilderTest
  • org.apache.mahout.df.mapreduce.partial.Step1Mapper
    • 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.