randomforest.cpp

   1 //
   2 //  randomforest.cpp
   3 //  Mothur
   4 //
   5 //  Created by Sarah Westcott on 10/2/12.
   6 //  Copyright (c) 2012 Schloss Lab. All rights reserved.
   7 //
   8
   9 #include "randomforest.hpp"
  10
  11 /***********************************************************************/
  12
  13 RandomForest::RandomForest(const vector <vector<int> > dataSet,
  14                            const int numDecisionTrees,
  15                            const string treeSplitCriterion = "gainratio",
  16                            const bool doPruning = false,
  17                            const float pruneAggressiveness = 0.9,
  18                            const bool discardHighErrorTrees = true,
  19                            const float highErrorTreeDiscardThreshold = 0.4,
  20                            const string optimumFeatureSubsetSelectionCriteria = "log2",
  21                            const float featureStandardDeviationThreshold = 0.0)
  22             : Forest(dataSet, numDecisionTrees, treeSplitCriterion, doPruning, pruneAggressiveness, discardHighErrorTrees, highErrorTreeDiscardThreshold, optimumFeatureSubsetSelectionCriteria, featureStandardDeviationThreshold) {
  23     m = MothurOut::getInstance();
  24 }
  25
  26 /***********************************************************************/
  27 // DONE
  28 int RandomForest::calcForrestErrorRate() {
  29     try {
  30         int numCorrect = 0;
  31         for (map<int, vector<int> >::iterator it = globalOutOfBagEstimates.begin(); it != globalOutOfBagEstimates.end(); it++) {
  32
  33             if (m->control_pressed) { return 0; }
  34
  35             int indexOfSample = it->first;
  36             vector<int> predictedOutComes = it->second;
  37             vector<int>::iterator maxPredictedOutComeIterator = max_element(predictedOutComes.begin(), predictedOutComes.end());
  38             int majorityVotedOutcome = (int)(maxPredictedOutComeIterator - predictedOutComes.begin());
  39             int realOutcome = dataSet[indexOfSample][numFeatures];
  40
  41             if (majorityVotedOutcome == realOutcome) { numCorrect++; }
  42         }
  43
  44         // TODO: save or return forrestErrorRate for future use;
  45         double forrestErrorRate = 1 - ((double)numCorrect / (double)globalOutOfBagEstimates.size());
  46
  47         m->mothurOut("numCorrect = " + toString(numCorrect)+ "\n");
  48         m->mothurOut("forrestErrorRate = " + toString(forrestErrorRate)+ "\n");
  49
  50         return 0;
  51     }
  52         catch(exception& e) {
  53                 m->errorOut(e, "RandomForest", "calcForrestErrorRate");
  54                 exit(1);
  55         }
  56 }
  57 /***********************************************************************/
  58
  59 int RandomForest::printConfusionMatrix(map<int, string> intToTreatmentMap) {
  60     try {
  61         int numGroups = intToTreatmentMap.size();
  62         vector<vector<int> > cm(numGroups, vector<int>(numGroups, 0));
  63
  64         for (map<int, vector<int> >::iterator it = globalOutOfBagEstimates.begin(); it != globalOutOfBagEstimates.end(); it++) {
  65
  66             if (m->control_pressed) { return 0; }
  67
  68             int indexOfSample = it->first; //key
  69             vector<int> predictedOutComes = it->second; //value, vector of all predicted classes
  70             vector<int>::iterator maxPredictedOutComeIterator = max_element(predictedOutComes.begin(), predictedOutComes.end());
  71             int majorityVotedOutcome = (int)(maxPredictedOutComeIterator - predictedOutComes.begin());
  72             int realOutcome = dataSet[indexOfSample][numFeatures];
  73             cm[realOutcome][majorityVotedOutcome] = cm[realOutcome][majorityVotedOutcome] + 1;
  74         }
  75
  76         vector<int> fw;
  77         for (int w = 0; w <numGroups; w++) {
  78             fw.push_back(intToTreatmentMap[w].length());
  79         }
  80
  81         m->mothurOut("confusion matrix:\n\t\t");
  82         for (int k = 0; k < numGroups; k++) {
  83             //m->mothurOut(intToTreatmentMap[k] + "\t");
  84             cout << setw(fw[k]) << intToTreatmentMap[k] << "\t";
  85         }
  86         for (int i = 0; i < numGroups; i++) {
  87             cout << "\n" << setw(fw[i]) << intToTreatmentMap[i] << "\t";
  88             //m->mothurOut("\n" + intToTreatmentMap[i] + "\t");
  89             if (m->control_pressed) { return 0; }
  90             for (int j = 0; j < numGroups; j++) {
  91                 //m->mothurOut(toString(cm[i][j]) + "\t");
  92                 cout << setw(fw[i]) << cm[i][j] << "\t";
  93             }
  94         }
  95         //m->mothurOut("\n");
  96         cout << "\n";
  97
  98         return 0;
  99     }
 100
 101     catch(exception& e) {
 102                 m->errorOut(e, "RandomForest", "printConfusionMatrix");
 103                 exit(1);
 104         }
 105 }
 106
 107 /***********************************************************************/
 108
 109 int RandomForest::getMissclassifications(string filename, map<int, string> intToTreatmentMap, vector<string> names) {
 110     try {
 111         ofstream out;
 112         m->openOutputFile(filename, out);
 113         out <<"Sample\tRF classification\tActual classification\n";
 114         for (map<int, vector<int> >::iterator it = globalOutOfBagEstimates.begin(); it != globalOutOfBagEstimates.end(); it++) {
 115
 116             if (m->control_pressed) { return 0; }
 117
 118             int indexOfSample = it->first;
 119             vector<int> predictedOutComes = it->second;
 120             vector<int>::iterator maxPredictedOutComeIterator = max_element(predictedOutComes.begin(), predictedOutComes.end());
 121             int majorityVotedOutcome = (int)(maxPredictedOutComeIterator - predictedOutComes.begin());
 122             int realOutcome = dataSet[indexOfSample][numFeatures];
 123
 124             if (majorityVotedOutcome != realOutcome) {
 125                 out << names[indexOfSample] << "\t" << intToTreatmentMap[majorityVotedOutcome] << "\t" << intToTreatmentMap[realOutcome] << endl;
 126
 127             }
 128         }
 129
 130         out.close();
 131         return 0;
 132     }
 133         catch(exception& e) {
 134                 m->errorOut(e, "RandomForest", "getMissclassifications");
 135                 exit(1);
 136         }
 137 }
 138
 139 /***********************************************************************/
 140 int RandomForest::calcForrestVariableImportance(string filename) {
 141     try {
 142
 143         // follow the link: http://en.wikipedia.org/wiki/Dynamic_cast
 144         //if you are going to dynamically cast, aren't you undoing the advantage of abstraction. Why abstract at all?
 145         //could cause maintenance issues later if other types of Abstract decison trees are created that cannot be cast as a decision tree.
 146         for (int i = 0; i < decisionTrees.size(); i++) {
 147             if (m->control_pressed) { return 0; }
 148
 149             DecisionTree* decisionTree = dynamic_cast<DecisionTree*>(decisionTrees[i]);
 150
 151             for (int j = 0; j < numFeatures; j++) {
 152                 globalVariableImportanceList[j] += (double)decisionTree->variableImportanceList[j];
 153             }
 154         }
 155
 156         for (int i = 0;  i < numFeatures; i++) {
 157             globalVariableImportanceList[i] /= (double)numDecisionTrees;
 158         }
 159
 160         vector< pair<int, double> > globalVariableRanks;
 161         for (int i = 0; i < globalVariableImportanceList.size(); i++) {
 162             //cout << "[" << i << ',' << globalVariableImportanceList[i] << "], ";
 163             if (globalVariableImportanceList[i] > 0) {
 164                 pair<int, double> globalVariableRank(0, 0.0);
 165                 globalVariableRank.first = i;
 166                 globalVariableRank.second = globalVariableImportanceList[i];
 167                 globalVariableRanks.push_back(globalVariableRank);
 168             }
 169         }
 170
 171 //        for (int i = 0; i < globalVariableRanks.size(); i++) {
 172 //            cout << m->currentBinLabels[(int)globalVariableRanks[i][0]] << '\t' << globalVariableImportanceList[globalVariableRanks[i][0]] << endl;
 173 //        }
 174
 175
 176         VariableRankDescendingSorterDouble variableRankDescendingSorter;
 177         sort(globalVariableRanks.begin(), globalVariableRanks.end(), variableRankDescendingSorter);
 178
 179         ofstream out;
 180         m->openOutputFile(filename, out);
 181         out <<"OTU\tMean decrease accuracy\n";
 182         for (int i = 0; i < globalVariableRanks.size(); i++) {
 183             out << m->currentBinLabels[(int)globalVariableRanks[i].first] << '\t' << globalVariableImportanceList[globalVariableRanks[i].first] << endl;
 184         }
 185         out.close();
 186         return 0;
 187     }
 188         catch(exception& e) {
 189                 m->errorOut(e, "RandomForest", "calcForrestVariableImportance");
 190                 exit(1);
 191         }
 192 }
 193 /***********************************************************************/
 194 int RandomForest::populateDecisionTrees() {
 195     try {
 196
 197         vector<double> errorRateImprovements;
 198
 199         for (int i = 0; i < numDecisionTrees; i++) {
 200
 201             if (m->control_pressed) { return 0; }
 202             if (((i+1) % 100) == 0) {  m->mothurOut("Creating " + toString(i+1) + " (th) Decision tree\n");  }
 203
 204             // TODO: need to first fix if we are going to use pointer based system or anything else
 205             DecisionTree* decisionTree = new DecisionTree(dataSet, globalDiscardedFeatureIndices, OptimumFeatureSubsetSelector(optimumFeatureSubsetSelectionCriteria), treeSplitCriterion, featureStandardDeviationThreshold);
 206
 207             if (m->debug && doPruning) {
 208                 m->mothurOut("Before pruning\n");
 209                 decisionTree->printTree(decisionTree->rootNode, "ROOT");
 210             }
 211
 212             int numCorrect;
 213             double treeErrorRate;
 214
 215             decisionTree->calcTreeErrorRate(numCorrect, treeErrorRate);
 216             double prePrunedErrorRate = treeErrorRate;
 217
 218             if (m->debug) {
 219                 m->mothurOut("treeErrorRate: " + toString(treeErrorRate) + " numCorrect: " + toString(numCorrect) + "\n");
 220             }
 221
 222             if (doPruning) {
 223                 decisionTree->pruneTree(pruneAggressiveness);
 224                 if (m->debug) {
 225                     m->mothurOut("After pruning\n");
 226                     decisionTree->printTree(decisionTree->rootNode, "ROOT");
 227                 }
 228                 decisionTree->calcTreeErrorRate(numCorrect, treeErrorRate);
 229             }
 230             double postPrunedErrorRate = treeErrorRate;
 231
 232
 233             decisionTree->calcTreeVariableImportanceAndError(numCorrect, treeErrorRate);
 234             double errorRateImprovement = (prePrunedErrorRate - postPrunedErrorRate) / prePrunedErrorRate;
 235
 236             if (m->debug) {
 237                 m->mothurOut("treeErrorRate: " + toString(treeErrorRate) + " numCorrect: " + toString(numCorrect) + "\n");
 238                 if (doPruning) {
 239                     m->mothurOut("errorRateImprovement: " + toString(errorRateImprovement) + "\n");
 240                 }
 241             }
 242
 243
 244             if (discardHighErrorTrees) {
 245                 if (treeErrorRate < highErrorTreeDiscardThreshold) {
 246                     updateGlobalOutOfBagEstimates(decisionTree);
 247                     decisionTree->purgeDataSetsFromTree();
 248                     decisionTrees.push_back(decisionTree);
 249                     if (doPruning) {
 250                         errorRateImprovements.push_back(errorRateImprovement);
 251                     }
 252                 } else {
 253                     delete decisionTree;
 254                 }
 255             } else {
 256                 updateGlobalOutOfBagEstimates(decisionTree);
 257                 decisionTree->purgeDataSetsFromTree();
 258                 decisionTrees.push_back(decisionTree);
 259                 if (doPruning) {
 260                     errorRateImprovements.push_back(errorRateImprovement);
 261                 }
 262             }
 263         }
 264
 265         double avgErrorRateImprovement = -1.0;
 266         if (errorRateImprovements.size() > 0) {
 267             avgErrorRateImprovement = accumulate(errorRateImprovements.begin(), errorRateImprovements.end(), 0.0);
 268 //            cout << "Total " << avgErrorRateImprovement << " size " << errorRateImprovements.size() << endl;
 269             avgErrorRateImprovement /= errorRateImprovements.size();
 270         }
 271
 272         if (m->debug && doPruning) {
 273             m->mothurOut("avgErrorRateImprovement:" + toString(avgErrorRateImprovement) + "\n");
 274         }
 275         // m->mothurOut("globalOutOfBagEstimates = " + toStringVectorMap(globalOutOfBagEstimates)+ "\n");
 276
 277
 278         return 0;
 279     }
 280     catch(exception& e) {
 281         m->errorOut(e, "RandomForest", "populateDecisionTrees");
 282         exit(1);
 283     }
 284 }
 285 /***********************************************************************/
 286 // TODO: need to finalize bettween reference and pointer for DecisionTree [partially solved]
 287 // DONE: make this pure virtual in superclass
 288 // DONE
 289 int RandomForest::updateGlobalOutOfBagEstimates(DecisionTree* decisionTree) {
 290     try {
 291         for (map<int, int>::iterator it = decisionTree->outOfBagEstimates.begin(); it != decisionTree->outOfBagEstimates.end(); it++) {
 292
 293             if (m->control_pressed) { return 0; }
 294
 295             int indexOfSample = it->first;
 296             int predictedOutcomeOfSample = it->second;
 297
 298             if (globalOutOfBagEstimates.count(indexOfSample) == 0) {
 299                 globalOutOfBagEstimates[indexOfSample] = vector<int>(decisionTree->numOutputClasses, 0);
 300             };
 301
 302             globalOutOfBagEstimates[indexOfSample][predictedOutcomeOfSample] += 1;
 303         }
 304         return 0;
 305     }
 306     catch(exception& e) {
 307         m->errorOut(e, "RandomForest", "updateGlobalOutOfBagEstimates");
 308         exit(1);
 309     }
 310 }
 311 /***********************************************************************/
 312
 313