LassoRegression.cpp

Go to the documentation of this file.

//===========================================================================
/*!
 * 
 *
 * \brief       LASSO Regression
 * 
 * This program demonstrates LASSO regression for the identification
 * of sparse coefficient vectors.
 * 
 * 
 *
 * \author      T. Glasmachers
 * \date        2013
 *
 *
 * \par Copyright 1995-2017 Shark Development Team
 * 
 * <BR><HR>
 * This file is part of Shark.
 * <http://shark-ml.org/>
 * 
 * Shark is free software: you can redistribute it and/or modify
 * it under the terms of the GNU Lesser General Public License as published 
 * by the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 * 
 * Shark is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU Lesser General Public License for more details.
 * 
 * You should have received a copy of the GNU Lesser General Public License
 * along with Shark.  If not, see <http://www.gnu.org/licenses/>.
 *
 */
//===========================================================================

#include <shark/Data/DataDistribution.h>
#include <shark/ObjectiveFunctions/Loss/SquaredLoss.h>
#include <shark/Algorithms/Trainers/LassoRegression.h>

#include <iostream>
#include <fstream>

using namespace shark;
using namespace std;


class TestProblem : public LabeledDataDistribution<RealVector, RealVector>
{
public:
    TestProblem(size_t informative, size_t nnz, size_t dim)
    : m_informative(informative)
    , m_nnz(nnz)
    , m_dim(dim)
    { }


    void draw(RealVector& input, RealVector& label) const
    {
        input.resize(m_dim);
        input.clear();
        label.resize(1);

        // we have one informative component per example
        double g = random::gauss(random::globalRng);
        size_t i = random::discrete(random::globalRng, std::size_t(0), m_informative-1);
        input(i) = g;
        label(0) = g;

        // the rest is non-informative
        for (size_t n=1; n<m_nnz; n++)
        {
            size_t i = random::discrete(random::globalRng, m_informative, m_dim-1);
            input(i) = random::gauss(random::globalRng);
        }
    }

protected:
    size_t m_informative;
    size_t m_nnz;
    size_t m_dim;
};


int main(int argc, char** argv)
{
    // Define a test problem with 10 out of 1000 informative
    // components. Each instance contains one informative and
    // 49 noise components. 10000 instances are drawn.
    TestProblem prob(10, 50, 1000);
    cout << "generating 100000 points ..." << flush;
    RegressionDataset data = prob.generateDataset(100000);
    cout << " done." << endl;

    // Set the regularization parameter.
    // For this problem the LASSO method identifies the correct
    // subset of 10 informative coefficients for a large range
    // of parameter values.
    double lambda = 1.0;

    // trainer and model
    LinearModel<> model;
    LassoRegression<> trainer(lambda);

    // train the model
    cout << "LASSO training ..." << flush;
    trainer.train(model, data);
    cout << " done." << endl;

    // check non-zero coefficients
    RealMatrix m = model.matrix();
    size_t nnz = 0;
    size_t correct = 0;
    size_t wrong = 0;
    for (size_t j=0; j<m.size2(); j++)
    {
        if (m(0, j) != 0.0)
        {
            nnz++;
            if (j < 10) correct++;
            else wrong++;
        }
    }
    cout << "solution statistics:" << endl;
    cout << "  number of non-zero coefficients: " << nnz << endl;
    cout << "  correctly identified coefficients: " << correct << endl;
    cout << "  wrongly identified coefficients: " << wrong << endl;
}