DecisionTreeClusterNode.cpp

#define GRT_DLL_EXPORTS
#include "DecisionTreeClusterNode.h"

GRT_BEGIN_NAMESPACE

//Register the DecisionTreeClusterNode module with the Node base class
RegisterNode< DecisionTreeClusterNode > DecisionTreeClusterNode::registerModule("DecisionTreeClusterNode");

DecisionTreeClusterNode::DecisionTreeClusterNode() : DecisionTreeNode("DecisionTreeClusterNode") {
 clear();
}

DecisionTreeClusterNode::~DecisionTreeClusterNode(){
 clear();
}

bool DecisionTreeClusterNode::predict_(VectorFloat &x) {

 if( x[ featureIndex ] >= threshold ) return true;

 return false;
}

bool DecisionTreeClusterNode::clear(){

 //Call the base class clear function
 DecisionTreeNode::clear();

 featureIndex = 0;
 threshold = 0;

 return true;
}

bool DecisionTreeClusterNode::print() const{

 std::ostringstream stream;

 if( getModel( stream ) ){
 std::cout << stream.str();
 return true;
 }

 return false;
}

bool DecisionTreeClusterNode::computeFeatureWeights( VectorFloat &weights ) const{

 if( isLeafNode ){ //If we reach a leaf node, no weight update needed
 return true;
 }
 
 if( featureIndex >= weights.getSize() ){ //Feature index is out of bounds
 warningLog << __GRT_LOG__ << " Feature index is greater than weights Vector size!" << std::endl;
 return false;
 }else{
 weights[ featureIndex ]++;
 }
 
 if( leftChild ){ //Recursively compute the weights for the left child
 leftChild->computeFeatureWeights( weights );
 }
 if( rightChild ){ //Recursively compute the weights for the right child
 rightChild->computeFeatureWeights( weights );
 }

 return true;
}

bool DecisionTreeClusterNode::computeLeafNodeWeights( MatrixFloat &weights ) const{

 if( isLeafNode ){ //If we reach a leaf node, there is nothing to do
 return true;
 }

 if( featureIndex >= weights.getNumCols() ){ //Feature index is out of bounds
 warningLog << __GRT_LOG__ << " Feature index is greater than weights Vector size!" << std::endl;
 return false;
 }

 if( leftChild ){ //Recursively compute the weights for the left child until we reach the node above a leaf node
 if( leftChild->getIsLeafNode() ){
 if( classProbabilities.getSize() != weights.getNumRows() ){
 warningLog << __GRT_LOG__ << " The number of rows in the weights matrix does not match the class probabilities Vector size!" << std::endl;
 return false;
 }
 for(UINT i=0; i<classProbabilities.getSize(); i++){
 weights[ i ][ featureIndex ] += classProbabilities[ i ];
 }
 
 } leftChild->computeLeafNodeWeights( weights );
 }
 if( rightChild ){ //Recursively compute the weights for the right child until we reach the node above a leaf node
 if( rightChild->getIsLeafNode() ){
 if( classProbabilities.getSize() != weights.getNumRows() ){
 warningLog << __GRT_LOG__ << " The number of rows in the weights matrix does not match the class probabilities Vector size!" << std::endl;
 return false;
 }
 for(UINT i=0; i<classProbabilities.getSize(); i++){
 weights[ i ][ featureIndex ] += classProbabilities[ i ];
 }
 } rightChild->computeLeafNodeWeights( weights );
 }

 return true;
}

bool DecisionTreeClusterNode::getModel( std::ostream &stream ) const{

 std::string tab = "";
 for(UINT i=0; i<depth; i++) tab += "\t";

 stream << tab << "depth: " << depth;
 stream << " nodeSize: " << nodeSize;
 stream << " featureIndex: " << featureIndex;
 stream << " threshold " << threshold;
 stream << " isLeafNode: " << isLeafNode << std::endl;

 stream << tab << "ClassProbabilities: ";
 for(UINT i=0; i<classProbabilities.getSize(); i++){
 stream << classProbabilities[i] << "\t";
 }
 stream << std::endl;

 if( leftChild != NULL ){
 stream << tab << "LeftChild: " << std::endl;
 leftChild->getModel( stream );
 }

 if( rightChild != NULL ){
 stream << tab << "RightChild: " << std::endl;
 rightChild->getModel( stream );
 }

 return true;
}

Node* DecisionTreeClusterNode::deepCopy() const{

 DecisionTreeClusterNode *node = new DecisionTreeClusterNode;

 if( node == NULL ){
 return NULL;
 }

 //Copy this node into the node
 node->depth = depth;
 node->isLeafNode = isLeafNode;
 node->nodeID = nodeID;
 node->predictedNodeID = predictedNodeID;
 node->nodeSize = nodeSize;
 node->featureIndex = featureIndex;
 node->threshold = threshold;
 node->classProbabilities = classProbabilities;

 //Recursively deep copy the left child
 if( leftChild ){
 node->leftChild = leftChild->deepCopy();
 node->leftChild->setParent( node );
 }

 //Recursively deep copy the right child
 if( rightChild ){
 node->rightChild = rightChild->deepCopy();
 node->rightChild->setParent( node );
 }

 return dynamic_cast< Node* >( node );
}

UINT DecisionTreeClusterNode::getFeatureIndex() const{
 return featureIndex;
}

Float DecisionTreeClusterNode::getThreshold() const{
 return threshold;
}

bool DecisionTreeClusterNode::set(const UINT nodeSize,const UINT featureIndex,const Float threshold,const VectorFloat &classProbabilities){
 this->nodeSize = nodeSize;
 this->featureIndex = featureIndex;
 this->threshold = threshold;
 this->classProbabilities = classProbabilities;
 return true;
}

bool DecisionTreeClusterNode::computeBestSplitBestIterativeSplit( const UINT &numSplittingSteps, const ClassificationData &trainingData, const Vector< UINT > &features, const Vector< UINT > &classLabels, UINT &featureIndex, Float &minError ){
 return computeSplit( numSplittingSteps, trainingData, features, classLabels, featureIndex, minError);
}

bool DecisionTreeClusterNode::computeBestSplitBestRandomSplit( const UINT &numSplittingSteps, const ClassificationData &trainingData, const Vector< UINT > &features, const Vector< UINT > &classLabels, UINT &featureIndex, Float &minError ){
 return computeSplit( numSplittingSteps, trainingData, features, classLabels, featureIndex, minError);
}

bool DecisionTreeClusterNode::computeSplit( const UINT &numSplittingSteps, const ClassificationData &trainingData, const Vector< UINT > &features, const Vector< UINT > &classLabels, UINT &featureIndex, Float &minError ){

 const UINT M = trainingData.getNumSamples();
 const UINT N = features.getSize();
 const UINT K = classLabels.getSize();

 if( N == 0 ) return false;
 if( K == 0 ) return false;

 minError = grt_numeric_limits< Float >::max();
 Random random;
 UINT bestFeatureIndex = 0;
 Float bestThreshold = 0;
 Float error = 0;
 Vector< UINT > groupIndex(M);
 Vector< MinMax > ranges = trainingData.getRanges();
 MatrixDouble data(M,1); //This will store our temporary data for each dimension

 //Randomly select which features we want to use
 UINT numRandomFeatures = numSplittingSteps > N ? N : numSplittingSteps;
 Vector< UINT > randomFeatures = random.getRandomSubset( 0, N, numRandomFeatures );

 //Loop over each random feature and try and find the best split point
 for(UINT n=0; n<numRandomFeatures; n++){

 featureIndex = features[ randomFeatures[n] ];

 //Use the data in this feature dimension to create a sum dataset
 for(UINT i=0; i<M; i++){
 data[i][0] = trainingData[i][featureIndex];
 }

 if( computeError( trainingData, data, classLabels, ranges, groupIndex, featureIndex, threshold, error ) ){
 //Store the best threshold and feature index
 if( error < minError ){
 minError = error;
 bestThreshold = threshold;
 bestFeatureIndex = featureIndex;
 }
 }
 }

 //Set the best feature index that will be returned to the DecisionTree that called this function
 featureIndex = bestFeatureIndex;

 //Store the node size, feature index, best threshold and class probabilities for this node
 set( M, featureIndex, bestThreshold, trainingData.getClassProbabilities(classLabels) );

 return true;
}

bool DecisionTreeClusterNode::computeError( const ClassificationData &trainingData, MatrixFloat &data, const Vector< UINT > &classLabels, Vector< MinMax > ranges, Vector< UINT > groupIndex, const UINT featureIndex, Float &threshold, Float &error ){

 error = 0;
 threshold = 0;

 const UINT M = trainingData.getNumSamples();
 const UINT K = (UINT)classLabels.size();

 Float giniIndexL = 0;
 Float giniIndexR = 0;
 Float weightL = 0;
 Float weightR = 0;
 VectorFloat groupCounter(2,0);
 MatrixFloat classProbabilities(K,2);

 //Use this data to train a KMeans cluster with 2 clusters
 KMeans kmeans;
 kmeans.setNumClusters( 2 );
 kmeans.setComputeTheta( true );
 kmeans.setMinChange( 1.0e-5 );
 kmeans.setMinNumEpochs( 1 );
 kmeans.setMaxNumEpochs( 100 );

 //Disable the logging to clean things up
 kmeans.setTrainingLoggingEnabled( false );

 if( !kmeans.train_( data ) ){
 errorLog << __GRT_LOG__ << " Failed to train KMeans model for feature: " << featureIndex << std::endl;
 return false;
 }

 //Set the split threshold as the mid point between the two clusters
 const MatrixFloat &clusters = kmeans.getClusters();
 threshold = 0;
 for(UINT i=0; i<clusters.getNumRows(); i++){
 threshold += clusters[i][0];
 }
 threshold /= clusters.getNumRows();

 //Iterate over each sample and work out if it should be in the lhs (0) or rhs (1) group based on the current threshold
 groupCounter[0] = groupCounter[1] = 0;
 classProbabilities.setAllValues(0);
 for(UINT i=0; i<M; i++){
 groupIndex[i] = trainingData[ i ][ featureIndex ] >= threshold ? 1 : 0;
 groupCounter[ groupIndex[i] ]++;
 classProbabilities[ getClassLabelIndexValue(trainingData[i].getClassLabel(),classLabels) ][ groupIndex[i] ]++;
 }

 //Compute the class probabilities for the lhs group and rhs group
 for(UINT k=0; k<K; k++){
 classProbabilities[k][0] = groupCounter[0]>0 ? classProbabilities[k][0]/groupCounter[0] : 0;
 classProbabilities[k][1] = groupCounter[1]>0 ? classProbabilities[k][1]/groupCounter[1] : 0;
 }

 //Compute the Gini index for the lhs and rhs groups
 giniIndexL = giniIndexR = 0;
 for(UINT k=0; k<K; k++){
 giniIndexL += classProbabilities[k][0] * (1.0-classProbabilities[k][0]);
 giniIndexR += classProbabilities[k][1] * (1.0-classProbabilities[k][1]);
 }
 weightL = groupCounter[0]/M;
 weightR = groupCounter[1]/M;
 error = (giniIndexL*weightL) + (giniIndexR*weightR);

 return true;
}

bool DecisionTreeClusterNode::saveParametersToFile( std::fstream &file ) const{

 if( !file.is_open() )
 {
 errorLog << __GRT_LOG__ << " File is not open!" << std::endl;
 return false;
 }

 //Save the DecisionTreeNode parameters
 if( !DecisionTreeNode::saveParametersToFile( file ) ){
 errorLog << __GRT_LOG__ << " Failed to save DecisionTreeNode parameters to file!" << std::endl;
 return false;
 }

 //Save the custom DecisionTreeThresholdNode parameters
 file << "FeatureIndex: " << featureIndex << std::endl;
 file << "Threshold: " << threshold << std::endl;

 return true;
}

bool DecisionTreeClusterNode::loadParametersFromFile( std::fstream &file ){

 if(!file.is_open())
 {
 errorLog << __GRT_LOG__ << " File is not open!" << std::endl;
 return false;
 }

 //Load the DecisionTreeNode parameters
 if( !DecisionTreeNode::loadParametersFromFile( file ) ){
 errorLog << __GRT_LOG__ << " Failed to load DecisionTreeNode parameters from file!" << std::endl;
 return false;
 }

 std::string word;
 //Load the custom DecisionTreeThresholdNode Parameters
 file >> word;
 if( word != "FeatureIndex:" ){
 errorLog << __GRT_LOG__ << " Failed to find FeatureIndex header!" << std::endl;
 return false;
 }
 file >> featureIndex;

 file >> word;
 if( word != "Threshold:" ){
 errorLog << __GRT_LOG__ << " Failed to find Threshold header!" << std::endl;
 return false;
 }
 file >> threshold;

 return true;
}

GRT_END_NAMESPACE