example_kmeansnn.cpp

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

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

Para executar o programa:

./example_kmeansnn

Retornando por consola:

Working Kmeans over ID: 0
┌──────────────────────────────┐
│    Kmeans                    │
│    IterationConf init data   │
├──────────────────────────────┤
│        MinError: 1.0000e-05  │
│         MaxIter: 1e+04       │
└──────────────────────────────┘
┌──────────────────────────────┐
│            Iter: 22          │
│       LastError: 0.0000e+00  │
│ Elapsed time(s): 6.5e-05     │
└──────────────────────────────┘

Working Kmeans over ID: 1
┌──────────────────────────────┐
│    Kmeans                    │
│    IterationConf init data   │
├──────────────────────────────┤
│        MinError: 1.0000e-05  │
│         MaxIter: 1e+04       │
└──────────────────────────────┘
┌──────────────────────────────┐
│            Iter: 23          │
│       LastError: 0.0000e+00  │
│ Elapsed time(s): 6.3e-05     │
└──────────────────────────────┘

Metrics of training:
╔═══════════════════════════════════╗
║    ClassificationMetrics data     ║
╠═══════════════════════════════════╣
║        Threshold: 0.5             ║
║          Samples: 1000            ║
╠═══════════════════════════════════╣
║ Pred.[0]/Act.[*]: 485     12      ║
║ Pred.[1]/Act.[*]: 15      488     ║
╠═══════════════════════════════════╣
║         Accuracy: 97.3  %         ║
║        Precision: 97.02 %         ║
║           Recall: 97.6  %         ║
╠═══════════════════════════════════╣
║           FScore: 97.31 %         ║
╚═══════════════════════════════════╝

Metrics of testing:
╔═══════════════════════════════════╗
║    ClassificationMetrics data     ║
╠═══════════════════════════════════╣
║        Threshold: 0.5             ║
║          Samples: 1000            ║
╠═══════════════════════════════════╣
║ Pred.[0]/Act.[*]: 472     17      ║
║ Pred.[1]/Act.[*]: 28      483     ║
╠═══════════════════════════════════╣
║         Accuracy: 95.5  %         ║
║        Precision: 94.52 %         ║
║           Recall: 96.6  %         ║
╠═══════════════════════════════════╣
║           FScore: 95.55 %         ║
╚═══════════════════════════════════╝

Classification - data

Classification - training

Classification - testing

Código example_kmeansnn.cpp:

 
#include <iostream>
#include <Pds/Ra>
#include <Pds/Ml>
 
    
int main(void)
{
    Pds::ClassificationMetrics Metrics;
    Pds::IterationConf Conf;
    Conf.SetMaxIter(10000); Conf.Show=true;
    Pds::Octave::XLabel="x_1";
    Pds::Octave::YLabel="x_2";
    
    unsigned int K=5; // Number of clusters
    unsigned int L=1000; // Numer of 1s or 0s
    Pds::Matrix X;
    Pds::Vector Y,Yp;
    
    // Generating training data
    Pds::DataSet::Classification::LoadDataYinYang(L,X,Y);
    
    Pds::Octave::Plot::PointsX2DY(X,Y,"testando.m","example_kmeansnn_data.png");
 
    /********* Training  *********/
    
    // Create KmeansNn
    Pds::KmeansNn KmNn(Conf,X,Y,K);
    
    // Predict "training" data with Knn=7
    Yp=KmNn.Predict(X,7);
    
    // Metrics of training
    Metrics = Pds::ClassificationMetrics::Calculate(0.5,Yp,Y);
    Metrics.Print("\nMetrics of training:\n");
    
    Pds::Octave::Plot::PointsX2DY(X,Yp,"testando.m","example_kmeansnn_training.png");
    
    /********* Testing  *********/
    
    // Generating testing data
    Pds::DataSet::Classification::LoadDataYinYang(L,X,Y);
    
    // Predict testing data with Knn=7
    Yp=KmNn.Predict(X,7);
    
    // Metrics testing
    Metrics = Pds::ClassificationMetrics::Calculate(0.5,Yp,Y);
    Metrics.Print("\nMetrics of testing:\n");
    
    Pds::Octave::Plot::PointsX2DY(X,Yp,"testando.m","example_kmeansnn_testing.png");
    return 0;
}