forest.cpp

   1 //
   2 //  forest.cpp
   3 //  Mothur
   4 //
   5 //  Created by Kathryn Iverson on 10/26/12.
   6 //  Copyright (c) 2012 Schloss Lab. All rights reserved.
   7 //
   8
   9 #include "forest.h"
  10
  11 /***********************************************************************/
  12 Forest::Forest(const std::vector < std::vector<int> > dataSet,
  13                                            const int numDecisionTrees,
  14                                            const string treeSplitCriterion = "informationGain")
  15 : dataSet(dataSet),
  16 numDecisionTrees(numDecisionTrees),
  17 numSamples((int)dataSet.size()),
  18 numFeatures((int)(dataSet[0].size() - 1)),
  19 globalVariableImportanceList(numFeatures, 0),
  20 treeSplitCriterion(treeSplitCriterion) {
  21     m = MothurOut::getInstance();
  22     globalDiscardedFeatureIndices = getGlobalDiscardedFeatureIndices();
  23     // TODO: double check if the implemenatation of 'globalOutOfBagEstimates' is correct
  24 }
  25
  26 /***********************************************************************/
  27
  28 vector<int> Forest::getGlobalDiscardedFeatureIndices() {
  29     try {
  30         //vector<int> globalDiscardedFeatureIndices;
  31         //globalDiscardedFeatureIndices.push_back(1);
  32
  33         // calculate feature vectors
  34         vector< vector<int> > featureVectors(numFeatures, vector<int>(numSamples, 0) );
  35         for (int i = 0; i < numSamples; i++) {
  36             if (m->control_pressed) { return globalDiscardedFeatureIndices; }
  37             for (int j = 0; j < numFeatures; j++) { featureVectors[j][i] = dataSet[i][j]; }
  38         }
  39
  40         for (int i = 0; i < featureVectors.size(); i++) {
  41             if (m->control_pressed) { return globalDiscardedFeatureIndices; }
  42             double standardDeviation = m->getStandardDeviation(featureVectors[i]);
  43             if (standardDeviation <= 0){ globalDiscardedFeatureIndices.push_back(i); }
  44         }
  45
  46         if (m->debug) {
  47             m->mothurOut("number of global discarded features:  " + toString(globalDiscardedFeatureIndices.size())+ "\n");
  48             m->mothurOut("total features: " + toString(featureVectors.size())+ "\n");
  49         }
  50
  51         return globalDiscardedFeatureIndices;
  52     }
  53         catch(exception& e) {
  54                 m->errorOut(e, "Forest", "getGlobalDiscardedFeatureIndices");
  55                 exit(1);
  56         }
  57 }
  58
  59 /***********************************************************************/
  60