X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=abstractdecisiontree.hpp;h=cc238b4687e57c6cab840f5ec31ce10adbc8a655;hb=541bab1dac00688b4c3a8c4a95ab464412663c50;hp=3445db4a511e8705ed5b0bed14626eb869ce5043;hpb=90708fe9701e3827e477c82fb3652539c3bf2a0d;p=mothur.git

diff --git a/abstractdecisiontree.hpp b/abstractdecisiontree.hpp
index 3445db4..cc238b4 100755
--- a/abstractdecisiontree.hpp
+++ b/abstractdecisiontree.hpp
@@ -6,8 +6,8 @@
 //  Copyright (c) 2012 Schloss Lab. All rights reserved.
 //
 
-#ifndef rrf_fs_prototype_abstractdecisiontree_hpp
-#define rrf_fs_prototype_abstractdecisiontree_hpp
+#ifndef RF_ABSTRACTDECISIONTREE_HPP
+#define RF_ABSTRACTDECISIONTREE_HPP
 
 #include "mothurout.h"
 #include "macros.h"
@@ -17,14 +17,22 @@
 
 /**************************************************************************************************/
 
+struct IntPairVectorSorter{
+    bool operator() (const pair<int, int>& firstPair, const pair<int, int>& secondPair) {
+        return firstPair.first < secondPair.first;
+    }
+};
+
+/**************************************************************************************************/
+
 class AbstractDecisionTree{
   
 public:
   
-    AbstractDecisionTree(vector<vector<int> >baseDataSet, 
-                       vector<int> globalDiscardedFeatureIndices, 
-                       OptimumFeatureSubsetSelector optimumFeatureSubsetSelector, 
-                       string treeSplitCriterion);    
+    AbstractDecisionTree(vector<vector<int> >& baseDataSet,
+                           vector<int> globalDiscardedFeatureIndices, 
+                           OptimumFeatureSubsetSelector optimumFeatureSubsetSelector, 
+                           string treeSplitCriterion);    
     virtual ~AbstractDecisionTree(){}
     
   
@@ -32,23 +40,29 @@ protected:
   
     virtual int createBootStrappedSamples();
     virtual int getMinEntropyOfFeature(vector<int> featureVector, vector<int> outputVector, double& minEntropy, int& featureSplitValue, double& intrinsicValue);
-    virtual int getBestSplitAndMinEntropy(vector< vector<int> > featureOutputPairs, vector<int> splitPoints, double& minEntropy, int& minEntropyIndex, double& relatedIntrinsicValue);
+        virtual int getBestSplitAndMinEntropy(vector< pair<int, int> > featureOutputPairs, vector<int> splitPoints, double& minEntropy, int& minEntropyIndex, double& relatedIntrinsicValue);
     virtual double calcIntrinsicValue(int numLessThanValueAtSplitPoint, int numGreaterThanValueAtSplitPoint, int numSamples);
-    virtual double calcSplitEntropy(vector< vector<int> > featureOutputPairs, int splitIndex, int numOutputClasses, bool);
+    virtual double calcSplitEntropy(vector< pair<int, int> > featureOutputPairs, int splitIndex, int numOutputClasses, bool);
+
     virtual int getSplitPopulation(RFTreeNode* node, vector< vector<int> >& leftChildSamples, vector< vector<int> >& rightChildSamples);
     virtual bool checkIfAlreadyClassified(RFTreeNode* treeNode, int& outputClass);
 
-    vector< vector<int> > baseDataSet;
+    vector< vector<int> >& baseDataSet;
     int numSamples;
     int numFeatures;
     int numOutputClasses;
     vector<int> outputClasses;
+    
     vector< vector<int> > bootstrappedTrainingSamples;
     vector<int> bootstrappedTrainingSampleIndices;
     vector< vector<int> > bootstrappedTestSamples;
     vector<int> bootstrappedTestSampleIndices;
     
+    vector<vector<int> > testSampleFeatureVectors;
+    
     RFTreeNode* rootNode;
+    int nodeIdCount;
+    map<int, int> nodeMisclassificationCounts;
     vector<int> globalDiscardedFeatureIndices;
     int optimumFeatureSubsetSize;
     string treeSplitCriterion;