example_logisticregression_ig.cpp

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

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

Para executar o programa:

./example_logisticregression_ig

Retornando por consola:

    W_ig: 0.4812        0       -1
 Cost(W): 0.15893202856795

-------------------------
   IterationConf data    
-------------------------
    Alpha: 0.1
 MinError: 1e-07
  MaxIter: 1000
-------------------------
iter:   1000 Alpha: 0.020986616658128 last_error: 0.0074522609311292 
-------------------------
    Alpha: 0.020986616658128
LastError: 0.0074522609311292
 LastIter: 1000
-------------------------
    W_ig: -0.73261682797554     -0.63672966050461       2.1557255295481
 Cost(W): 0.30983446123018

Classification - data

Classification - ortogonal

Classification - gradient

Código example_logisticregression_ig.cpp:

 
#include <iostream>
#include <Pds/Ra>
#include <Pds/Ml>
        
int main(void)
{
    Pds::IterationConf Conf;    
    Conf.Show=true; Conf.SetMinError(1e-07); Conf.SetAlpha(0.01); Conf.SetLambda(0.01); Conf.SetGamma(0.1);
    Pds::Ra::Randomize();
    
    // Generating data
    //Pds::Matrix X(Pds::Ra::TextFormat,"../test/hard-data-x.txt");
    //Pds::Vector Y(Pds::Ra::TextFormat,"../test/hard-data-y.txt");
    unsigned int L=1000;
    Pds::Matrix X;
    Pds::Vector Y;
    Pds::DataSet::Classification::LoadDataBand(L,X,Y);
    
    Pds::Octave::XLabel="x_1";
    Pds::Octave::YLabel="x_2";
    Pds::Octave::Plot::PointsX2DY(X,Y,"testando.m","example_logisticregression_ig_data.png");
 
    Pds::Vector Yp;
    Pds::Vector W;
    
    // Create W using training data
    
    // Método ortogonal
    W=Pds::LogisticModel::FittingOrtogonalIG(X,Y,1);
 
    W.T().Print("    W_ig: ");
    std::cout<< " Cost(W): "<<Pds::LogisticModel::CostInformationGain(W,X,Y)<<"\n\n";    
    
    Yp=Pds::LogisticModel::Classify(W,X);
    
    Pds::Octave::Plot::PointsX2DY(X,Yp,"testando.m","example_logisticregression_ig_ortogonal.png");
    
    // Método de gradiente
    W=Pds::LogisticModel::GetW0MeanMethod(X);   
    W=Pds::LogisticModel::FittingGradientIG(Conf,X,Y,W);
    
    W.T().Print("    W_ig: ");
    std::cout<< " Cost(W): "<<Pds::LogisticModel::CostInformationGain(W,X,Y)<<"\n"; 
    
    Yp=Pds::LogisticModel::Classify(W,X);
    
    Pds::Octave::Plot::PointsX2DY(X,Yp,"testando_gradient.m","example_logisticregression_ig_gradient.png");
    
    return 0;
}