doxygen_page

# A single Shogun program
This documents classififier_libsvm_probabilities.cpp.

Cool thing is we can use standard markdown to write the documentation while the source code can be executed. Another advantage is that not all source code is pasted at once. Rather, we can write little paragraphs in between, even including math.

## Include and initialisation
First, let's include the needed header files and initialise Shogun.
However, if this is not documented in this doxypage (one might not want to do that as it is standard), the code still is in the source file and thus can be executed and tested.

### Header includes
\snippet classifier_libsvm_probabilities.cpp	 headers
### Initialising Shogun
\snippet classifier_libsvm_probabilities.cpp	 init_shogun



## SVM
First, we define some constants
\snippet classifier_libsvm_probabilities.cpp	 define_constants
Then, training an SVM, even using math \f$ f(x)=\sum_{i=1}^N \alpha_i k(x,x_i) \f$ blablabla
\snippet classifier_libsvm_probabilities.cpp	 train_svm

## Fitting probabilities to SVM output
Convert scores to calibrated probabilities  by fitting a sigmoid function
using the method described in Lin, H., Lin, C., and Weng,  R. (2007). A note
on Platt's probabilistic outputs for support vector machines.
See CBinaryLabels documentation for details
\snippet classifier_libsvm_probabilities.cpp	 scores_to_probabilities

source_file.cpp

//! [headers]
#include <shogun/base/init.h>
#include <shogun/features/DenseFeatures.h>
#include <shogun/kernel/LinearKernel.h>
#include <shogun/labels/BinaryLabels.h>
#include <shogun/classifier/svm/LibSVM.h>
#include <iostream>
//! [headers]


using namespace shogun;

void gen_rand_data(SGMatrix<float64_t> features, SGVector<float64_t> labels, float64_t distance)
{
    index_t num_samples=labels.vlen;
    index_t dimensions=features.num_rows;
    for (int32_t i=0; i<num_samples; i++)
    {
        if (i<num_samples/2)
        {
            labels[i]=-1.0;
            for(int32_t j=0; j<dimensions; j++)
                features(j,i)=CMath::random(0.0,1.0)+distance;
        }
        else
        {
            labels[i]=1.0;
            for(int32_t j=0; j<dimensions; j++)
                features(j,i)=CMath::random(0.0,1.0)-distance;
        }
    }
    labels.display_vector("labels");
    std::cout<<std::endl;
    features.display_matrix("features");
    std::cout<<std::endl;
}

int main(int argc, char** argv)
{
//! [init_shogun]
    init_shogun_with_defaults();
//! [init_shogun]

//! [define_constants]
    const float64_t svm_C=10;

    index_t num_samples=20;
    index_t dimensions=2;
    float64_t dist=0.5;
//! [define_constants]

    SGMatrix<float64_t> featureMatrix(dimensions,num_samples);
    SGVector<float64_t> labelVector(num_samples);
    gen_rand_data(featureMatrix,labelVector,dist);

    CLabels* labels=new CBinaryLabels(labelVector);

    CDenseFeatures<float64_t>* features=new CDenseFeatures<float64_t>();
    SG_REF(features);
    features->set_feature_matrix(featureMatrix);

    CLinearKernel* kernel=new CLinearKernel();
    SG_REF(kernel);
    kernel->init(features, features);

    //! [train_svm]
    CLibSVM* svm=new CLibSVM(svm_C,kernel, labels);
    SG_REF(svm);
    svm->train();
	//! [train_svm]


	//! [scores_to_probabilities]
    CBinaryLabels* out_labels=CLabelsFactory::to_binary(svm->apply());
    out_labels->scores_to_probabilities();
	//! [scores_to_probabilities]

    for (int32_t i=0; i<num_samples; i++)
    {
        SG_SPRINT("out[%d]=%f (%f)\n", i, out_labels->get_label(i),
            out_labels->get_value(i));
    }

    SG_UNREF(out_labels);
    SG_UNREF(kernel);
    SG_UNREF(features);
    SG_UNREF(svm);

    exit_shogun();

    return 0;
}