source: trunk/test/ncc_test.cc @ 1050

// $Id: ncc_test.cc 1050 2008-02-07 18:47:34Z peter $

/*
  Copyright (C) 2006 Jari Häkkinen, Markus Ringnér

  This file is part of the yat library, http://trac.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 2 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 this program; if not, write to the Free Software
  Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
  02111-1307, USA.
*/

#include "yat/classifier/IGP.h"
#include "yat/classifier/Kernel_MEV.h"
#include "yat/classifier/KernelLookup.h"
#include "yat/classifier/MatrixLookup.h"
#include "yat/classifier/MatrixLookupWeighted.h"
#include "yat/classifier/NCC.h"
#include "yat/classifier/PolynomialKernelFunction.h"
#include "yat/classifier/Target.h"
#include "yat/utility/matrix.h"
#include "yat/statistics/euclidean_distance.h"
#include "yat/statistics/pearson_distance.h"
#include "yat/utility/utility.h"

#include <cassert>
#include <fstream>
#include <iostream>
#include <stdexcept>
#include <sstream>
#include <string>
#include <limits>
#include <cmath>

using namespace theplu::yat;

double deviation(const utility::matrix& a, const utility::matrix& b) {
  double sl=0;
  for (size_t i=0; i<a.rows(); i++){
    for (size_t j=0; j<a.columns(); j++){
      sl += fabs(a(i,j)-b(i,j));
    }
  }
  sl /= (a.columns()*a.rows());
  return sl;
}

int main(const int argc,const char* argv[])
{  

  std::ostream* error;
  if (argc>1 && argv[1]==std::string("-v"))
    error = &std::cerr;
  else {
    error = new std::ofstream("/dev/null");
    if (argc>1)
      std::cout << "ncc_test -v : for printing extra information\n";
  }
  *error << "testing ncc" << std::endl;
  bool ok = true;

  /////////////////////////////////////////////
  // First test of constructor and training  
  /////////////////////////////////////////////
  classifier::MatrixLookup ml(4,4);
  std::vector<std::string> vec(4, "pos");
  vec[3]="bjds";
  classifier::Target target(vec);
  classifier::NCC<statistics::EuclideanDistance> ncctmp(ml,target);
  *error << "training...\n";
  ncctmp.train();
  *error << "done\n";

  /////////////////////////////////////////////
  // A test of predictions using unweighted data
  /////////////////////////////////////////////
  utility::matrix data1(3,4);
  for(size_t i=0;i<3;i++) {
    data1(i,0)=3-i;
    data1(i,1)=5-i;
    data1(i,2)=i+1;
    data1(i,3)=i+3;
  }
  std::vector<std::string> vec1(4, "pos");
  vec1[0]="neg";
  vec1[1]="neg";

  classifier::MatrixLookup ml1(data1);
  classifier::Target target1(vec1);

  classifier::NCC<statistics::EuclideanDistance> ncc1(ml1,target1);
  ncc1.train();
  utility::matrix prediction1;
  ncc1.predict(ml1,prediction1);
  double slack_bound=2e-7;
  utility::matrix result1(2,4);
  result1(0,0)=result1(0,1)=result1(1,2)=result1(1,3)=sqrt(3.0);
  result1(0,2)=result1(0,3)=result1(1,0)=result1(1,1)=sqrt(11.0);
  double slack = deviation(prediction1,result1); 
  if (slack > slack_bound || std::isnan(slack)){
    *error << "Difference to expected prediction too large\n";
    *error << "slack: " << slack << std::endl;
    *error << "expected less than " << slack_bound << std::endl;
    ok = false;
  }

  //////////////////////////////////////////////////////////////////////////
  // A test of predictions using unweighted training and weighted test data
  //////////////////////////////////////////////////////////////////////////
  utility::matrix weights1(3,4,1.0);
  weights1(0,0)=weights1(1,1)=weights1(2,2)=weights1(1,3)=0.0;
  classifier::MatrixLookupWeighted mlw1(data1,weights1);
  ncc1.predict(mlw1,prediction1); 
  result1(0,2)=result1(0,3)=result1(1,0)=result1(1,1)=sqrt(15.0);
  slack = deviation(prediction1,result1);
  if (slack > slack_bound || std::isnan(slack)){
    *error << "Difference to expected prediction too large\n";
    *error << "slack: " << slack << std::endl;
    *error << "expected less than " << slack_bound << std::endl;
    ok = false;
  }

  

  //////////////////////////////////////////////////////////////////////////
  // A test of predictions using Sorlie data
  //////////////////////////////////////////////////////////////////////////
  std::ifstream is("data/sorlie_centroid_data.txt");
  utility::matrix data(is,'\t');
  is.close();

  is.open("data/sorlie_centroid_classes.txt");
  classifier::Target targets(is);
  is.close();

  // Generate weight matrix with 0 for missing values and 1 for others.
  utility::matrix weights(data.rows(),data.columns(),0.0);
  utility::nan(data,weights);
      
  classifier::MatrixLookupWeighted dataviewweighted(data,weights);
  classifier::NCC<statistics::EuclideanDistance> ncc(dataviewweighted,targets);
  *error << "training...\n";
  ncc.train();

  // Comparing the centroids to stored result
  is.open("data/sorlie_centroids.txt");
  utility::matrix centroids(is);
  is.close();

  if(centroids.rows() != ncc.centroids().rows() ||
     centroids.columns() != ncc.centroids().columns()) {
    *error << "Error in the dimensionality of centroids\n";
    *error << "Nof rows: " << centroids.rows() << " expected: " 
           << ncc.centroids().rows() << std::endl;
    *error << "Nof columns: " << centroids.columns() << " expected: " 
           << ncc.centroids().columns() << std::endl;
  }

  slack = deviation(centroids,ncc.centroids());
  if (slack > slack_bound || std::isnan(slack)){
    *error << "Difference to stored centroids too large\n";
    *error << "slack: " << slack << std::endl;
    *error << "expected less than " << slack_bound << std::endl;
    ok = false;
  }

  *error << "...predicting...\n";
  utility::matrix prediction;
  ncc.predict(dataviewweighted,prediction);
  
  // Comparing the prediction to stored result
  is.open("data/sorlie_centroid_predictions.txt");
  utility::matrix result(is,'\t');
  is.close();

  slack = deviation(result,prediction);
  if (slack > slack_bound || std::isnan(slack)){
    *error << "Difference to stored prediction too large\n";
    *error << "slack: " << slack << std::endl;
    *error << "expected less than " << slack_bound << std::endl;
    ok = false;
  }
  *error << "done\n";

  //////////////////////////////////////////////////////////////////////////
  // Testing rejection of KernelLookups
  //////////////////////////////////////////////////////////////////////////
  classifier::PolynomialKernelFunction kf;
  classifier::Kernel_MEV kernel(ml,kf);
  classifier::DataLookup2D* dl_kernel = new classifier::KernelLookup(kernel); 
  try {
    ok=0; // should catch error here
    *error << ncc.make_classifier(*dl_kernel,target) << std::endl;
  }
  catch (std::runtime_error) {    
    *error << "caught expected bad cast runtime_error" << std::endl;
    ok=1;
  }
  try {
    ok=0; // should catch error here
    ncc.predict(*dl_kernel,prediction); 
  }
  catch (std::runtime_error) {    
    *error << "caught expected bad cast runtime_error" << std::endl;
    ok=1;
  }
  delete dl_kernel;
  
  if(ok)
    *error << "OK" << std::endl;

  if (error!=&std::cerr)
    delete error;

  if(ok) 
    return 0;
  return -1;  
}