KfoldTimeSeriesData::KfoldTimeSeriesData(GRT::LabelledTimeSeriesClassificationData inputDataset):
inputDataset(inputDataset){
crossValidationSetup = false;
crossValidationIndexs.clear();
kFoldValue = 0;
classTracker = inputDataset.getClassTracker();
numDimensions = inputDataset.getNumDimensions();
UINT totalNumSamples = inputDataset.getNumSamples();
}
HRESULT GesturesAnalyzer::Init()
{
m_DTWTrainingData.loadDatasetFromFile("TrainingData\\DTWTrainingData.txt");
m_ANBCTrainingData.loadDatasetFromFile("TrainingData\\ANBCTrainingData.txt");
m_HandTrainingData.loadDatasetFromFile("TrainingData\\HandTrainingData.txt");
m_hand.enableScaling( true );
m_hand.enableNullRejection ( true );
m_hand.setNullRejectionCoeff( 5 );
m_anbc.enableScaling( true );
m_anbc.enableNullRejection( true );
m_anbc.setNullRejectionCoeff( 1 );
m_dtw.addPreProcessingModule( GRT::MovingAverageFilter( 5, m_DTWTrainingData.getNumDimensions() ) );
m_dtw.setClassifier( GRT::DTW() );
//m_dtw.getClassifier()->enableNullRejection( true );
//m_dtw.enableScaling(true);
////m_dtw.enableZNormalization(true);
//m_dtw.enableNullRejection( true );
//m_dtw.enableTrimTrainingData(true,0.1,90);
//m_dtw.setOffsetTimeseriesUsingFirstSample( true );
GRT::LabelledTimeSeriesClassificationData testData = m_DTWTrainingData.partition( 80, true );
if( !m_anbc.train( m_ANBCTrainingData ) ){
std::cout << "Failed to train ANBC classifier!\n";
return E_FAIL;
}
if( !m_dtw.train( m_DTWTrainingData ) ){
std::cout << "Failed to train DTW classifier!\n";
return E_FAIL;
}
double accuracy = 0;
for(UINT i=0; i<testData.getNumSamples(); i++){
//Get the i'th test sample - this is a timeseries
UINT classLabel = testData[i].getClassLabel();
GRT::MatrixDouble timeseries = testData[i].getData();
//Perform a prediction using the classifier
if( !m_dtw.predict( timeseries ) ){
cout << "Failed to perform prediction for test sampel: " << i <<"\n";
return EXIT_FAILURE;
}
//Get the predicted class label
UINT predictedClassLabel = m_dtw.getPredictedClassLabel();
double maximumLikelihood = m_dtw.getMaximumLikelihood();
GRT::VectorDouble classLikelihoods = m_dtw.getClassLikelihoods();
GRT::VectorDouble classDistances = m_dtw.getClassDistances();
//Update the accuracy
if( classLabel == predictedClassLabel ) accuracy++;
cout << "TestSample: " << i << "\tClassLabel: " << classLabel << "\tPredictedClassLabel: " << predictedClassLabel << "\tMaximumLikelihood: " << maximumLikelihood << endl;
}
cout << "Test Accuracy: " << accuracy/double(testData.getNumSamples())*100.0 << "%" << endl;
if( !m_hand.train(m_HandTrainingData))
{
std::cout << "Failed to train Hand classifier!\n";
return E_FAIL;
}
return S_OK;
}