abstractdecisiontree.hpp

   1 //
   2 //  abstractdecisiontree.hpp
   3 //  rrf-fs-prototype
   4 //
   5 //  Created by Abu Zaher Faridee on 7/22/12.
   6 //  Copyright (c) 2012 Schloss Lab. All rights reserved.
   7 //
   8
   9 #ifndef RF_ABSTRACTDECISIONTREE_HPP
  10 #define RF_ABSTRACTDECISIONTREE_HPP
  11
  12 #include "mothurout.h"
  13 #include "macros.h"
  14 #include "rftreenode.hpp"
  15
  16 #define DEBUG_MODE
  17
  18 /**************************************************************************************************/
  19
  20 struct IntPairVectorSorter{
  21     bool operator() (const pair<int, int>& firstPair, const pair<int, int>& secondPair) {
  22         return firstPair.first < secondPair.first;
  23     }
  24 };
  25
  26 /**************************************************************************************************/
  27
  28 class AbstractDecisionTree{
  29
  30 public:
  31
  32     AbstractDecisionTree(vector<vector<int> >& baseDataSet,
  33                            vector<int> globalDiscardedFeatureIndices,
  34                            OptimumFeatureSubsetSelector optimumFeatureSubsetSelector,
  35                            string treeSplitCriterion);
  36     virtual ~AbstractDecisionTree(){}
  37
  38
  39 protected:
  40
  41     virtual int createBootStrappedSamples();
  42     virtual int getMinEntropyOfFeature(vector<int> featureVector, vector<int> outputVector, double& minEntropy, int& featureSplitValue, double& intrinsicValue);
  43         virtual int getBestSplitAndMinEntropy(vector< pair<int, int> > featureOutputPairs, vector<int> splitPoints, double& minEntropy, int& minEntropyIndex, double& relatedIntrinsicValue);
  44     virtual double calcIntrinsicValue(int numLessThanValueAtSplitPoint, int numGreaterThanValueAtSplitPoint, int numSamples);
  45     virtual double calcSplitEntropy(vector< pair<int, int> > featureOutputPairs, int splitIndex, int numOutputClasses, bool);
  46
  47     virtual int getSplitPopulation(RFTreeNode* node, vector< vector<int> >& leftChildSamples, vector< vector<int> >& rightChildSamples);
  48     virtual bool checkIfAlreadyClassified(RFTreeNode* treeNode, int& outputClass);
  49
  50     vector< vector<int> >& baseDataSet;
  51     int numSamples;
  52     int numFeatures;
  53     int numOutputClasses;
  54     vector<int> outputClasses;
  55
  56     vector< vector<int> > bootstrappedTrainingSamples;
  57     vector<int> bootstrappedTrainingSampleIndices;
  58     vector< vector<int> > bootstrappedTestSamples;
  59     vector<int> bootstrappedTestSampleIndices;
  60
  61     vector<vector<int> > testSampleFeatureVectors;
  62
  63     RFTreeNode* rootNode;
  64     int nodeIdCount;
  65     map<int, int> nodeMisclassificationCounts;
  66     vector<int> globalDiscardedFeatureIndices;
  67     int optimumFeatureSubsetSize;
  68     string treeSplitCriterion;
  69     MothurOut* m;
  70
  71 private:
  72
  73
  74 };
  75 /**************************************************************************************************/
  76
  77 #endif
  78
  79