/**
*@ test training model of classification algorithm.
*@ param classifyMethod classifymethod: "knn" , "naviebayes".
*@ param testdata: the primary trainging data.
*@ param testDataLabel: the class label of primary training data.
*@ param metric: distance function.
*@ param pivotsAndTrainModelFileName: the file to store the selected pivots information and training model.
*@ param testModelFileName: the file to store the training model.
*@ param status: determine the source of the test data. 0: TestDataFromTrainData; 1: TestDataFromTestData.
*@ param k: the number of nearest neighbors
*/
void CTest_Knn::TestModel(char *classifyMethod,vector<shared_ptr<CMetricData> > *testdata,vector<string> testDataLabel, CMetricDistance *metric,char *pivotsAndTrainModelFileName,char *testModelFileName,int status,int k,int splitRatio)
{
CDatasetInMetricSpace getTestData;
GetMetricData M_testdata;
if(status==0)
{
M_testdata=getTestData.getMetricTestData_fromTrainData(classifyMethod,testdata,testDataLabel,metric,pivotsAndTrainModelFileName,splitRatio);
pivotNumber=getTestData.pivotsNum;
}
else if(status==1)
{
M_testdata=getTestData.getMetricTestData_fromTestData(classifyMethod,testdata,testDataLabel,metric,pivotsAndTrainModelFileName);
pivotNumber=getTestData.pivotsNum;
}
showClassiciationResult(pivotsAndTrainModelFileName,M_testdata,testModelFileName,pivotNumber,k);
}
void Train(char *dataType,char *disfun,char *pivotSelectionMethod,int numPivot,char *classifyMethod,char *trainDataFileName,int initialSize,int dim,char *pivotsAndTrainModelFileName,int coordinate)
{
#ifdef __GNUC__
char *newTrainDataFileName="../../../SourceFiles/util/data/";
#endif
#ifdef _WIN32
char *newTrainDataFileName="./SourceFiles/util/data/";
#endif
joinCharArray(newTrainDataFileName,trainDataFileName);
#ifdef __GNUC__
char *newPivotsAndTrainModelFileName="../../../SourceFiles/util/result/";
#endif
#ifdef _WIN32
char *newPivotsAndTrainModelFileName="./SourceFiles/util/result/";
#endif
joinCharArray(newPivotsAndTrainModelFileName,pivotsAndTrainModelFileName);
vector<shared_ptr<CMetricData> > *traindata=0; //store the traindata
vector<string> trainDataLabel;
if(strcmp(dataType,"vector")==0)
{
traindata=CDoubleVector::loadData(newTrainDataFileName,initialSize,dim);
CReadLabel readTrainDataLabel;
trainDataLabel = readTrainDataLabel.loadLabel(newTrainDataFileName,initialSize,dataType);
}
else if(strcmp(dataType,"string")==0)
{
traindata=CStringObject::loadData(newTrainDataFileName,initialSize);
CReadLabel readTrainDataLabel;
trainDataLabel = readTrainDataLabel.loadLabel(newTrainDataFileName,initialSize,dataType);
}
else if(strcmp(dataType,"dna")==0)
{
traindata=CDNA_CLASSIFY::loadData(newTrainDataFileName,initialSize,dim);
CReadLabel readTrainDataLabel;
trainDataLabel = readTrainDataLabel.loadLabel(newTrainDataFileName,initialSize/57,dataType);
}
else if(strcmp(dataType,"time_series")==0)
{
traindata=CTimeSeries::loadData(newTrainDataFileName,initialSize,dim);
CReadLabel readTrainDataLabel;
trainDataLabel = readTrainDataLabel.loadLabel(newTrainDataFileName,initialSize,dataType);
}
else if(strcmp(dataType,"image")==0)
{
traindata=CImage::loadData(newTrainDataFileName,initialSize,dim);
}
else if(strcmp(dataType,"peptide")==0)
{
traindata=CPeptide::loadData(newTrainDataFileName,initialSize,dim);
}
CMetricDistance *metric=0;
if(strcmp(disfun,"EuclideanDistance")==0)
{
metric = new CEuclideanDistance;
}
else if(strcmp(disfun,"EditDistance")==0)
{
metric = new CEditDistance;
}
else if(strcmp(disfun,"TimeSeriesMetric")==0)
{
metric = new CTimeSeriesMetric;
}
else if(strcmp(disfun,"ImageMetric")==0)
{
metric = new CImageMetric;
}
else if(strcmp(disfun,"DNAMetric")==0)
{
metric = new CDNAMetric;
}
else if(strcmp(disfun,"PeptideMetric")==0)
{
metric = new CPeptideMetric;
}
else if(strcmp(disfun,"LInfinityDistance")==0)
{
metric = new CLInfinityDistance;
}
CountedMetric *cmetric=new CountedMetric(metric);
CPivotSelectionMethod *pivotselectionmethod=0;
if(strcmp(pivotSelectionMethod,"fft")==0)
{
pivotselectionmethod = new CFFTPivotSelectionMethod;
}
else if(strcmp(pivotSelectionMethod,"pcaonfft")==0)
{
pivotselectionmethod = new CPCAOnFFT(2);
}
else if(strcmp(pivotSelectionMethod,"incremental")==0)
{
pivotselectionmethod = new CIncrementalSelection(1,2);
}
if(numPivot > initialSize)
{
cout << "The pivot number larger than the traindata size. Please reset the number of pivot !" << endl;
exit(0);
}
if(strcmp(classifyMethod,"knn")==0)
{
CTrain_Knn train_knn;
train_knn.TrainModel(classifyMethod,traindata,trainDataLabel,cmetric,pivotselectionmethod,numPivot,newPivotsAndTrainModelFileName,dim,coordinate);
}
else if(strcmp(classifyMethod,"knnCrossValid")==0)
{
CTrain_Knn train_knn;
train_knn.TrainModelUseCrossValidation(classifyMethod,traindata,trainDataLabel,cmetric,pivotselectionmethod,numPivot,newPivotsAndTrainModelFileName,dim,coordinate);
}
else if(strcmp(classifyMethod,"naviebayes")==0)
{
CTrain_NavieBayes train_naviebayes;
train_naviebayes.TrainModel(classifyMethod,traindata,trainDataLabel,cmetric,pivotselectionmethod,numPivot,newPivotsAndTrainModelFileName,newTrainDataFileName,dim,coordinate);
}
else if(strcmp(classifyMethod,"c4.5")==0)
{
C4_5 c45;
c45.TrainModel(classifyMethod,traindata,trainDataLabel,cmetric,pivotselectionmethod,numPivot,newPivotsAndTrainModelFileName,newTrainDataFileName,dim,coordinate);
}
else if(strcmp(classifyMethod,"svm")==0)
{
CDatasetInMetricSpace getTrainData;
GetMetricData M_traindata;
M_traindata=getTrainData.getMetricTrainData(classifyMethod,traindata,trainDataLabel,cmetric,pivotselectionmethod,numPivot,newPivotsAndTrainModelFileName,dim,coordinate);
CTrain_SVM svm(M_traindata);
svm.TrainModel(newPivotsAndTrainModelFileName);
}
}
void Test(char *dataType,char *disfun,char *classifyMethod,char *testDataFileName,int finalSize,int dim,char *pivotsAndTrainModelFileName,char *testModelFileName,int status,int k,int splitRatio)
{
#ifdef __GNUC__
char *newTestDataFileName="../../../SourceFiles/util/data/";
#endif
#ifdef _WIN32
char *newTestDataFileName="./SourceFiles/util/data/";
#endif
joinCharArray(newTestDataFileName,testDataFileName);
#ifdef __GNUC__
char *newPivotsAndTrainModelFileName="../../../SourceFiles/util/result/";
#endif
#ifdef _WIN32
char *newPivotsAndTrainModelFileName="./SourceFiles/util/result/";
#endif
joinCharArray(newPivotsAndTrainModelFileName,pivotsAndTrainModelFileName);
#ifdef __GNUC__
char *newTestModelFileName="../../../SourceFiles/util/result/";
#endif
#ifdef _WIN32
char *newTestModelFileName="./SourceFiles/util/result/";
#endif
joinCharArray(newTestModelFileName,testModelFileName);
vector<shared_ptr<CMetricData> > *testdata=0; //store the testdata
vector<string> testDataLabel;
if(strcmp(dataType,"vector")==0)
{
testdata=CDoubleVector::loadData(newTestDataFileName,finalSize,dim);
CReadLabel readTestDataLabel;
testDataLabel=readTestDataLabel.loadLabel(newTestDataFileName,finalSize,dataType);
}
else if(strcmp(dataType,"string")==0)
{
testdata=CStringObject::loadData(newTestDataFileName,finalSize);
CReadLabel readTestDataLabel;
testDataLabel=readTestDataLabel.loadLabel(newTestDataFileName,finalSize,dataType);
}
else if(strcmp(dataType,"dna")==0)
{
testdata=CDNA_CLASSIFY::loadData(newTestDataFileName,finalSize,dim);
CReadLabel readTestDataLabel;
testDataLabel=readTestDataLabel.loadLabel(newTestDataFileName,finalSize/57,dataType);
}
else if(strcmp(dataType,"time_series")==0)
{
testdata=CTimeSeries::loadData(newTestDataFileName,finalSize,dim);
CReadLabel readTestDataLabel;
testDataLabel=readTestDataLabel.loadLabel(newTestDataFileName,finalSize,dataType);
}
else if(strcmp(dataType,"image")==0)
{
testdata=CImage::loadData(newTestDataFileName ,finalSize, dim);
}
else if(strcmp(dataType,"peptide")==0)
{
testdata=CPeptide::loadData(newTestDataFileName,finalSize,dim);
}
CMetricDistance *metric=0;
if(strcmp(disfun,"EuclideanDistance")==0)
{
metric = new CEuclideanDistance;
}
else if(strcmp(disfun,"EditDistance")==0)
{
metric = new CEditDistance;
}
else if(strcmp(disfun,"TimeSeriesMetric")==0)
{
metric = new CTimeSeriesMetric;
}
else if(strcmp(disfun,"ImageMetric")==0)
{
metric = new CImageMetric;
}
else if(strcmp(disfun,"DNAMetric")==0)
{
metric = new CDNAMetric;
}
else if(strcmp(disfun,"PeptideMetric")==0)
{
metric = new CPeptideMetric;
}
else if(strcmp(disfun,"LInfinityDistance")==0)
{
metric = new CLInfinityDistance;
}
CountedMetric *cmetric=new CountedMetric(metric);
if(strcmp(classifyMethod,"knn")==0)
{
CTest_Knn test_knn;
test_knn.TestModel(classifyMethod,testdata,testDataLabel,cmetric,newPivotsAndTrainModelFileName,newTestModelFileName,status,k,splitRatio);
}
else if(strcmp(classifyMethod,"naviebayes")==0)
{
CTest_NavieBayes test_naviebayes;
test_naviebayes.TestModel(classifyMethod,testdata,testDataLabel,cmetric,newPivotsAndTrainModelFileName,newTestModelFileName,status,splitRatio);
}
else if(strcmp(classifyMethod,"c4.5")==0)
{
CTest_C4_5 test_c45;
test_c45.TestModel(classifyMethod,testdata,testDataLabel,cmetric,newPivotsAndTrainModelFileName,newTestModelFileName,status,splitRatio);
}
else if(strcmp(classifyMethod,"svm")==0)
{
CDatasetInMetricSpace getTestData;
GetMetricData M_testdata;
if(status==0)
{
M_testdata=getTestData.getMetricTestData_fromTrainData(classifyMethod,testdata,testDataLabel,cmetric,newPivotsAndTrainModelFileName,splitRatio);
printf("\n\nEvaluation on training data (%d items):\n", M_testdata.metricData.size());
}
else if(status==1)
{
M_testdata=getTestData.getMetricTestData_fromTestData(classifyMethod,testdata,testDataLabel,cmetric,newPivotsAndTrainModelFileName);
printf("\nEvaluation on test data (%d items):\n", M_testdata.metricData.size());
}
CTest_SVM svm(M_testdata);
svm.TestModel(newPivotsAndTrainModelFileName,newTestModelFileName);
}
}
