example_logisticregression_ms.cpp

Programa para el testeo de las funciones.Para compilar o código example_logisticregression_ms.cpp:

g++ -static -o example_logisticregression_ms example_logisticregression_ms.cpp -lpdsmlmm -lpdsramm -lpdsspmm

Para executar o programa:

./example_logisticregression_ms

Retornando por consola:

W:
0.0031532109277748      4.5655191182942 4.6114943601509
Elapsed time: 4.766 ms

-----------------------------
 ClassificationMetrics data  
-----------------------------
Threshold       : 0.5
Samples         : 2000
----------------:------------
Pred.[0]/Act.[*]: 1006  0
Pred.[1]/Act.[*]: 3     991
----------------:------------
Accuracy        : 99.85%
Precision       : 99.7%
Recall          : 100%
----------------:------------
FScore          : 99.85%
-----------------------------

-----------------------------
 ClassificationMetrics data  
-----------------------------
Threshold       : 0.5
Samples         : 2000
----------------:------------
Pred.[0]/Act.[*]: 1009  2
Pred.[1]/Act.[*]: 2     987
----------------:------------
Accuracy        : 99.8%
Precision       : 99.8%
Recall          : 99.8%
----------------:------------
FScore          : 99.8%
-----------------------------

Classification - data

Classification - training

Classification - testing

Código example_logisticregression_ms.cpp:

 
#include <iostream>
#include <Pds/Ra>
#include <Pds/Ml>
        
int main(void)
{
    Pds::IterationConf Conf;    Conf.Show=true; Conf.SetMinError(1e-07);
    Pds::Vector Yp;
    Pds::ClassificationMetrics Metrics;
        
    // Generating data
    unsigned int L=5000;
    Pds::Matrix X;
    Pds::Vector Y;
    Pds::DataSet::Classification::LoadDataLine(L,X,Y);
    Pds::Vector W(X.Ncol()+1);
    
    // Split data set in {Training,Cross-validation,Test}
    Pds::DataSetBlock Dat=Pds::DataSet::Split(X,Y,6,2,2);
    
    Pds::Octave::XLabel="x_1";
    Pds::Octave::YLabel="x_2";
    Pds::Octave::Plot::PointsX2DY(Dat.Xtr,Dat.Ytr,"testando.m","example_logisticregression_ms_data.png");
    
    // Create W using training data
    W.Fill(0.1);    Conf.SetAlpha(0.1);
    Pds::Ra::Tic();
    W=Pds::LogisticModel::FittingLogitMeanSquare(Conf,Dat.Xtr,Dat.Ytr,0.01);
    W.T().Print("W:\n");
    Pds::Ra::Toc();
    
    // Testing W with testing data set
    Yp=Pds::LogisticModel::Classify(W,Dat.Xcv);
    
    // Metrics  
    Metrics = Pds::ClassificationMetrics::Optimal(Yp,Dat.Ycv);
    Metrics.Print("\n");
 
    Pds::Octave::Plot::PointsX2DYW( Dat.Xcv,Yp.Geq(Metrics.Threshold),W,
                                    "testando.m","example_logisticregression_ms_training.png");
    
    // Testing W with testing data set
    Yp=Pds::LogisticModel::Classify(W,Dat.Xtt);
    
    Metrics = Pds::ClassificationMetrics::Calculate(Metrics.Threshold,Yp,Dat.Ytt);
    Metrics.Print("\n");
    
    Pds::Octave::Plot::PointsX2DYW( Dat.Xtt,Yp.Geq(Metrics.Threshold),W,
                                    "testando.m","example_logisticregression_ms_testing.png");
 
    return 0;
}