source: branches/0.5-stable/test/ensemble_test.cc @ 2132

// $Id: ensemble_test.cc 2132 2009-12-24 01:13:05Z peter $

/*
  Copyright (C) 2006 Jari Häkkinen, Peter Johansson, Markus Ringnér
  Copyright (C) 2007 Jari Häkkinen, Peter Johansson
  Copyright (C) 2008 Jari Häkkinen, Peter Johansson, Markus Ringnér

  This file is part of the yat library, http://dev.thep.lu.se/yat

  The yat library is free software; you can redistribute it and/or
  modify it under the terms of the GNU General Public License as
  published by the Free Software Foundation; either version 3 of the
  License, or (at your option) any later version.

  The yat library 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
  General Public License for more details.

  You should have received a copy of the GNU General Public License
  along with yat. If not, see <http://www.gnu.org/licenses/>.
*/

#include "Suite.h"

#include "yat/utility/Matrix.h"
#include "yat/classifier/SubsetGenerator.h"
#include "yat/classifier/CrossValidationSampler.h"
#include "yat/classifier/EnsembleBuilder.h"
#include "yat/classifier/Kernel.h"
#include "yat/classifier/KernelLookup.h"
#include "yat/classifier/Kernel_SEV.h"
#include "yat/classifier/Kernel_MEV.h"
#include "yat/classifier/MatrixLookup.h"
#include "yat/classifier/MatrixLookupWeighted.h"
#include "yat/classifier/NCC.h"
#include "yat/classifier/PolynomialKernelFunction.h"
#include "yat/classifier/SVM.h"
#include "yat/statistics/AUC.h"
#include "yat/statistics/EuclideanDistance.h"

#include <cassert>
#include <fstream>
#include <iostream>
#include <cstdlib>
#include <limits>


int main(int argc, char* argv[])
{  
  using namespace theplu::yat;
  test::Suite suite(argc, argv);
  
  suite.err() << "testing ensemble" << std::endl;

  suite.err() << "loading data" << std::endl;
  std::ifstream is(test::filename("data/nm_data_centralized.txt").c_str());
  utility::Matrix data_core(is);
  is.close();

  suite.err() << "create MatrixLookup" << std::endl;
  classifier::MatrixLookup data(data_core);
  classifier::KernelFunction* kf = new classifier::PolynomialKernelFunction(); 
  suite.err() << "Building kernel" << std::endl;
  classifier::Kernel_SEV kernel(data,*kf);


  suite.err() << "load target" << std::endl;
  is.open(test::filename("data/nm_target_bin.txt").c_str());
  classifier::Target target(is);
  is.close();
  assert(data.columns()==target.size());

  {
    suite.err() << "create ensemble of ncc" << std::endl;
    classifier::NCC<statistics::EuclideanDistance> ncc;
    classifier::CrossValidationSampler sampler(target,3,3);
    classifier::SubsetGenerator<classifier::MatrixLookup> subdata(sampler,data);
    classifier::EnsembleBuilder<classifier::SupervisedClassifier,
      classifier::MatrixLookup> ensemble(ncc, data, sampler);
    suite.err() << "build ensemble" << std::endl;
    ensemble.build();
    std::vector<std::vector<statistics::Averager> > result;
    ensemble.predict(data, result);
  }

  {
    suite.err() << "create ensemble of ncc" << std::endl;
    classifier::MatrixLookupWeighted data_weighted(data);
    classifier::NCC<statistics::EuclideanDistance> ncc;
    classifier::CrossValidationSampler sampler(target,3,3);
    classifier::SubsetGenerator<classifier::MatrixLookupWeighted> 
      subdata(sampler,data_weighted);
    classifier::EnsembleBuilder<classifier::SupervisedClassifier,
      classifier::MatrixLookupWeighted> ensemble(ncc, data_weighted, sampler);
    suite.err() << "build ensemble" << std::endl;
    ensemble.build();
    std::vector<std::vector<statistics::Averager> > result;
    ensemble.predict(data_weighted, result);
  }

  suite.err() << "create KernelLookup" << std::endl;
  classifier::KernelLookup kernel_lookup(kernel);
  suite.err() << "create svm" << std::endl;
  classifier::SVM svm;
  suite.err() << "create Subsets" << std::endl;
  classifier::CrossValidationSampler sampler(target,3,3);
  classifier::SubsetGenerator<classifier::KernelLookup> cv(sampler,
                                                           kernel_lookup);

  suite.err() << "create ensemble" << std::endl;
  classifier::EnsembleBuilder<classifier::SVM, classifier::KernelLookup> 
    ensemble(svm, kernel_lookup, sampler);
  suite.err() << "build ensemble" << std::endl;
  ensemble.build();
  utility::Vector out(target.size(),0);
  for (size_t i = 0; i<out.size(); ++i) {
    out(i)=ensemble.validate()[0][i].mean(); 
  }
  statistics::AUC roc;
  suite.err() << roc.score(target,out) << std::endl;

  std::vector<std::vector<statistics::Averager> > result;
  ensemble.predict(kernel_lookup, result);
  for (size_t i = 0; i<result.size(); ++i) {
    for (size_t j=0; j<result[0].size(); ++j) {
      if (!suite.add(result[i][j].variance() > 0)) {
        suite.err() << "error: element " << i << " " << j << "\n";
        suite.err() << "expected finite prediction varince\n";
        suite.err() << "found: " << result[i][j].variance() << "\n";
      }
    } 
  }

  {
    suite.err() << "create ensemble" << std::endl;
    classifier::EnsembleBuilder<classifier::SVM, classifier::KernelLookup> 
      ensemble(svm, kernel_lookup, sampler);
    suite.err() << "test validate() before build()\n";
    ensemble.validate();
    std::vector<std::vector<statistics::Averager> > result;
    suite.err() << "test predict() before build()\n";
    ensemble.predict(kernel_lookup, result);
  }
  delete kf;

  return suite.return_value();
}