void TestAxisLabels()
{
CPlot2D *plot2D=new CPlot2D();
CDataSet dataSet;
CDataPoint point;
for (double x=-100.01;x<=100.0;x+=10.0) {
point.SetValues(x,x*x*x-x*x+x);
dataSet.AddDataPoint(point);
}
dataSet.SetDatasetColor(0.0,1.0,0.0);
dataSet.SetDatasetTitle("x^3-x^2+x");
plot2D->AddDataSet(dataSet);
plot2D->SetXAxisTitle("x axis");
plot2D->SetYAxisTitle("y=x^3-x^2+x axis");
plot2D->SetLeftFrameSize(100.0);
plot2D->OutputPostScriptPlot("test5_axis_labels.ps");
}
// Loads a new feature space and appends it to the multiset
void loadFeatureSpace( string filename, CDataSet< vSparseSample >& ds )
{
// Load the data
cout<<"Loading "<<filename<<"... "; cout.flush();
CDataSet<CSparseSample> d;
parseSparseFile( filename, d, ',', ',', '=' );
cout<<"Loaded "<<d.size()<<" samples; "<<nFeats( d )<<" features"<<endl;
// Append the Zhang dataset to the multiset
expand( ds, d );
cout<<" The combined dataset is now at "<<ds.size()<<" samples"<<endl;
}
void TestTrigFunctions()
{
CPlot2D *plot2D=new CPlot2D();
CDataSet dataSet;
CDataPoint point;
for (double x=-3.1515;x<=3.2;x+=0.1) {
point.SetValues(x,sin(x));
dataSet.AddDataPoint(point);
}
dataSet.SetDatasetColor(0.0,0.0,0.0);
dataSet.SetDatasetTitle("sin(x)");
plot2D->AddDataSet(dataSet);
CDataSet dataSet1;
for (double x=-3.1515;x<=3.2;x+=0.1) {
point.SetValues(x,cos(x));
dataSet1.AddDataPoint(point);
}
dataSet1.SetDatasetColor(1.0,0.0,0.0);
dataSet1.SetDatasetTitle("cos(x)");
plot2D->AddDataSet(dataSet1);
CDataSet dataSet2;
for (double x=-3.1515;x<=3.2;x+=0.1) {
point.SetValues(x,sin(x)*cos(x));
dataSet2.AddDataPoint(point);
}
dataSet2.SetDatasetColor(0.0,0.0,1.0);
dataSet2.SetDatasetTitle("sin(x)*cos(x)");
plot2D->AddDataSet(dataSet2);
plot2D->SetXAxisTitle("variable x");
plot2D->SetYAxisTitle("function");
plot2D->OutputPostScriptPlot("test1_trigfunctions.ps");
}
void CParticleSystem::SetParticlesFromData()
{
SingletonPointer<CDataManager> DataManager;
CDataSet * DataSet = DataManager->GetCurrentDataSet();
if (! DataSet)
{
return;
}
Particles.resize(DataSet->Size());
for (uint i = 0; i < DataSet->Size(); ++ i)
{
CDataRow const & Row = DataSet->GetRow(i);
double X = 0, Y = 0, Z = 0;
if (XField >= 0)
{
X = Row.GetFieldAsDouble(XField);
}
if (YField >= 0)
{
Y = Row.GetFieldAsDouble(YField);
}
if (ZField >= 0)
{
Z = Row.GetFieldAsDouble(ZField);
}
Particles[i].Data = vec3d(X, Y, Z);
Particles[i].Position = vec3d(X, Y, Z);
}
NormalizeParticlePositions();
UpdateColors();
Update();
}
CGreyModel::CGreyModel(CDataSet& his_data)
{
m_his_data.assign(his_data.begin(), his_data.end());
m_forecast_data.assign(his_data.begin(), his_data.end());
m_data_size = m_his_data.size();
cout<<"输入数据大小：m_data_size="<<m_data_size<<endl;
// print_his_data();
m_init_population_num = 5; //初始种群数目
m_max_population_num = 300; //最大种群数目
m_min_seed = 0; //最小种子数
m_max_seed = 5; //最大种子数
m_max_iteration_num = 100; //最大迭代次数
m_dimension = 2; //求解维数
m_nonlinear_parameter = 2; //非线性调制参数
m_init_stand_dev = 50.00; //初始标准差
m_final_stand_dev = 0.01; //最终标准差
m_a_min = -0.3; //a随机范围下限
m_a_max = 2; //a随机范围上限
m_u_min = 0.00; //u随机范围下限
m_u_max = 200.5; //u随机范围上限
m_curr_population_num = 0;
get_model();
forecast_data();
}
//----------------------------------------
int32_t CBratProcessExportAscii::WriteData()
{
int32_t result = BRATHL_SUCCESS;
//CTrace *p = CTrace::GetInstance();
//p->Tracer(1, "Writing data ...");
CObArray::iterator itDataSet;
CRecordSet* recordSet = NULL;
CDataSet* dataSet = m_product->GetDataSet();
for (itDataSet = dataSet->begin(); itDataSet != dataSet->end() ; itDataSet++)
{
recordSet = dataSet->GetRecordSet(itDataSet);
///bool firstRecord = (record == dataSet->front());
CExpressionValue exprValueSelect;
if (m_alwaysTrue == false)
{
recordSet->ExecuteExpression(m_select, m_recordName, exprValueSelect, m_product);
}
if (m_writeHeader)
{
for (uint32_t indexExpr = 0 ; indexExpr < m_fields.size() ; indexExpr++)
{
CExpressionValue exprValueData;
recordSet->ExecuteExpression(m_fields[indexExpr], m_recordName, exprValueData, m_product);
string colName = (m_names[indexExpr].empty() ? m_fields[indexExpr].AsString() : m_names[indexExpr]);
string unitStr;
unitStr.append(" (");
unitStr.append(m_units[indexExpr].AsString(false, true));
unitStr.append(") ");
if (m_units[indexExpr].IsDate() && !m_dateAsPeriod )
{
unitStr = "";
}
if (m_units[indexExpr].IsDate() && !m_units[indexExpr].HasDateRef() )
{
unitStr = "";
}
*m_outputFile << colName << unitStr << " " << exprValueData.GetDimensionsAsString() << "\t";
}
*m_outputFile << endl;
m_writeHeader = false;
}
if (m_alwaysTrue == false)
{
if (exprValueSelect.IsTrue() != 1)
{
//Data doesn't match SELECT options,
//-------------
continue;
//-------------
}
}
for (uint32_t indexExpr = 0 ; indexExpr < m_fields.size() ; indexExpr++)
{
CExpressionValue exprValueData;
recordSet->ExecuteExpression(m_fields[indexExpr], m_recordName, exprValueData, m_product);
*m_outputFile << exprValueData.AsString(m_units[indexExpr], m_formats[indexExpr], m_dateAsPeriod) << "\t";
}
*m_outputFile << endl;
}
//p->Tracer(1,"End writing data ...");
return result;
}
int main( int argc, char* argv[] )
{
// Display usage and/or parse the parameters
if( argc < 2 ) { cout<<"Usage: "<<argv[0]<<" <options file>"<<endl; return -1; }
CGOStrutParams params; params.load( argv[1] ); params.display();
typedef CKernelLoss<CSparseSample> CSparseKLoss;
CDataSet<vSparseSample>::binop_t fcker = CCompositeSparseKernel( false );
CDataSet<CSparseSample>::binop_t fker = CSparseKernel( true );
CDataSet<CSparseSample>::binop_t floss = CSparseKLoss( fker );
function< double( double, double ) > fioker = CProdJointKernel();
///////////////////////////////
CDataSet< vSparseSample > dsi;
// Load the datasets
loadFeatureSpace( "mousefunc/ge_zhang.sdat.gz", dsi );
loadFeatureSpace( "mousefunc/ge_su.sdat.gz", dsi );
loadFeatureSpace( "mousefunc/i_adj.sdat", dsi );
loadFeatureSpace( "mousefunc/dd_pfam.sdat", dsi );
loadFeatureSpace( "mousefunc/dd_inter.sdat", dsi );
loadFeatureSpace( "mousefunc/phylo.sdat", dsi );
shared_ptr< CFeatMap > pfm( new CFeatMap );
CDataSet<CSparseSample> dsoTr;
CDataSet<CSparseSample> dsoTe;
// Load the appropriate set of labels
if( GO::hasMF( params.ontology() ) )
loadOntologySpace( "molecular function", "mf", dsoTr, dsoTe, pfm );
if( GO::hasBP( params.ontology() ) )
loadOntologySpace( "biological process", "bp", dsoTr, dsoTe, pfm );
if( GO::hasCC( params.ontology() ) )
loadOntologySpace( "cellular compoennt", "cc", dsoTr, dsoTe, pfm );
if( dsoTr.size() < 1 || dsoTe.size() < 1 )
throw std::runtime_error( "No labels loaded" );
// Append hierarchy to the samples
cout<<"Training data output space has "<<dsoTr.size()<<" samples; "<<nFeats( dsoTr )<<" features"<<endl;
cout<<"Test data output space has "<<dsoTe.size()<<" samples; "<<nFeats( dsoTe )<<" features"<<endl;
// DEBUG
/* vector< unsigned int > v;
for( unsigned int i = 0; i < 400; i++ )
v.push_back( i );
dsoTr.subsample( v );*/
// END OF DEBUG
// Combine the data to form a joint output space
typedef CIODataSet<vSparseSample, CSparseSample> CIOSet;
shared_ptr< CIOSet > pdsio( new CIOSet(fcker, fker, floss, fioker) );
pdsio->addSets( dsi, dsoTr );
pdsio->addSets( dsi, dsoTe );
cout<<"Input space has "<<pdsio->sizeI()<<" samples"<<endl;
cout<<"Output space has "<<pdsio->sizeO()<<" samples, "<<nFeats( pdsio->getO() )<<" features"<<endl;
string ont;
if( GO::hasMF( params.ontology() ) ) ont += "mf";
if( GO::hasBP( params.ontology() ) ) ont += "bp";
if( GO::hasCC( params.ontology() ) ) ont += "cc";
string fnScores = "mfunc-score-" + params.alg_choice() + "-" + ont + ".sdat";
cout<<"Saving prediction scores to "<<fnScores<<endl;
// Re-split the data into training and test sets, both using the same output space now
// Retrieve the fold sizes
vector< unsigned int > foldSizes;
foldSizes.push_back( pdsio->sizeI() - dsoTe.size() );
foldSizes.push_back( dsoTe.size() );
// Create the ranges
virange_t vTrain, vTest;
splitCV( foldSizes, 1, vTrain, vTest );
display( vTrain, vTest );
// Perform the split
shared_ptr< CIOSet > pTrain;
shared_ptr< CIOSet > pTest;
cout<<"About to split"<<endl;
pdsio->splitTrainTest( vTrain, vTest, pTrain, pTest );
cout<<"Finished the split"<<endl;
pTrain->cache();
pTest->cache();
// Create and add the appropriate test classifier
shared_ptr< CClassifier<vSparseSample, CSparseSample> > pclsf = createClassifier<vSparseSample, CSparseSample>( params );
// Train and test the classifier
pclsf->train( pTrain );
vector< double > loss = pclsf->test( pTest );
// Report the results
double m = std::accumulate( loss.begin(), loss.end(), 0.0 ) /
static_cast<double>( loss.size() );
cout<<"Mean loss per test sample: "<<m<<endl;
pTrain->cacheIExternal( pTest->getI() );
predScores( *pclsf, pTest->getI(), fnScores );
return 0;
}
