example_ldc_2d.cpp

Programa para el testeo de las funciones.Para compilar o código example_ldc_2d.cpp:

g++ -static -o example_ldc_2d example_ldc_2d.cpp -lpdsmlmm -lpdsramm -lpdsspmm

Para executar o programa:

./example_ldc_2d

Retornando por consola:

----- LDC :
Sinv:
6.7407373495477   -0.74357495589715 
-0.74357495589715 11.553669016034   
Mu[0]:
0.99703793502286  0.045538886737149 
Mu[1]:
-0.47496384436618 0.83686241024232  
Mu[2]:
-0.47638727066386 -0.85447614834989 
C[0]:   -1.0986122886681
C[1]:   -1.0986122886681
C[2]:   -1.0986122886681
Label[0]:       0
Label[1]:       1
Label[2]:       2


----- LDC2 :
Sinv:
6.7407373495477   -0.74357495589715 
-0.74357495589715 11.553669016034   
Mu[0]:
0.99703793502286  0.045538886737149 
Mu[1]:
-0.47496384436618 0.83686241024232  
Mu[2]:
-0.47638727066386 -0.85447614834989 
C[0]:   -1.0986122886681
C[1]:   -1.0986122886681
C[2]:   -1.0986122886681
Label[0]:       0
Label[1]:       1
Label[2]:       2

Tarining data:

Training data

Confusion matrix:

Confusion matrix

Decision boundaries:

Regions

Código example_ldc_2d.cpp:

 
#include <iostream>
#include <Pds/Ra>
#include <Pds/Ml>
    
int main(void)
{
    unsigned int L=300;
    unsigned int K=3;
    Pds::Octave::MarkerSize=64;
    Pds::Octave::YLimits=true;
    Pds::Octave::XLimits=true;
    Pds::Octave::DAspect=true;
    
    Pds::Matrix X;
    Pds::Vector Y,Ys;
    Pds::ConfusionMatrix CM;
        
    // Load dataset of L samples an K gaussian normal.
    Pds::DataSet::Classification::LoadKGaussian(L,K,X,Y);
    
    // Plor de dataset X with label Y.
    Pds::Octave::Plot::ScatterX2DY(X,Y,"testando3d.m","example_ldc_2d_data.png");
 
    // Calculate the Ldc classifier.
    Pds::LDC Ldc(X,Y);
    Ldc.Print("----- LDC :\n");
 
    // Test of training
    Ys=Ldc.Predict(X);
 
    // Confusion matrix
    CM=Pds::ConfusionMatrix::Calculate(Ys,Y);
    Pds::Octave::Plot::ConfMatrix(CM.Data,Ldc.GetLabel(),"testando.m","example_ldc_2d_ConfMatrix.png");
 
    // Load new dataset Xt.
    Pds::Matrix Min=X.MinInCols();
    Pds::Matrix Max=X.MaxInCols();
    Pds::Matrix X1=Pds::Matrix(10*L,1); X1.FillRandU(Min.Get(0),Max.Get(0));
    Pds::Matrix X2=Pds::Matrix(10*L,1); X2.FillRandU(Min.Get(1),Max.Get(1));
    X=Pds::MergeHor({X1,X2});
    
    // Predict the datset Xt.
    Ys=Ldc.Predict(X);
    
    // Print the result of classification.
    Pds::Octave::Plot::ScatterX2DY(X,Ys,"testando3d.m","example_ldc_2d_class.png");
    
    Ldc.ExportFile("salida.txt");
    Pds::LDC Ldc2("salida.txt");
    Ldc.Print("----- LDC2 :\n");
 
    return 0;
}