source: trunk/yat/normalizer/Spearman.h @ 2154

#ifndef _theplu_yat_normalizer_spearman_
#define _theplu_yat_normalizer_spearman_

// $Id: Spearman.h 2154 2010-01-17 22:21:29Z peter $

/*
  Copyright (C) 2008 Jari Häkkinen, Peter Johansson
  Copyright (C) 2009 Peter Johansson

  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 "yat/utility/DataIterator.h"
#include "yat/utility/iterator_traits.h"
#include "yat/utility/sort_index.h"
#include "yat/utility/WeightIterator.h"

#include <boost/concept_check.hpp>

#include <algorithm>
#include <functional>
#include <vector>

namespace theplu {
namespace yat {
namespace normalizer {

  /**
     \brief Replace elements with normalized rank

     Each element x is replaced by \f$ \frac{\sum I(x_i-x) w_i}{\sum
     w_i} \f$ where \f$ I(x) = 1 \f$ for \f$ x>0 \f$, \f$I(x) = 0.5
     \f$ for \f$ x=0 \f$, and \f$ I(x) = 0 \f$ for \f$ x<0 \f$.

     \since New in yat 0.5
   */
  class Spearman
  {
  public:
    /**
       \brief default constructor
     */
    Spearman(void){}

    /**
       It is possible to centralize a range "in place"; it is
       permissible for the iterators \a first and \a result to be the
       same.

       Type Requirements:
       - RandomAccessIter1 must be a \random_access_iterator
       - RandomAccessIter2 must be a mutable \random_access_iterator

     */
    template<typename RandomAccessIter1, typename RandomAccessIter2>
    void operator()(RandomAccessIter1 first, RandomAccessIter1 last,
                    RandomAccessIter2 result) const
    {
      BOOST_CONCEPT_ASSERT((boost::RandomAccessIterator<RandomAccessIter1>));
      BOOST_CONCEPT_ASSERT((boost::Mutable_RandomAccessIterator<RandomAccessIter2>));
      typename utility::weighted_iterator_traits<RandomAccessIter1>::type tag;
      normalize(first, last, result, tag);
    }


  private:
    // unweighted version
    template<typename RandomAccessIter1, typename RandomAccessIter2>
    void normalize(RandomAccessIter1 first, RandomAccessIter1 last,
                   RandomAccessIter2 result, 
                   utility::unweighted_iterator_tag) const
    {
      std::vector<size_t> perm;
      utility::sort_index(first, last, perm);
      double n = perm.size();
      size_t i=0; 
      while ( i<perm.size() ) {
        size_t min_i = i;
        while (i<perm.size() && first[perm[i]]<=first[perm[min_i]])
          ++i;
        double res = static_cast<double>(i + min_i)/(2*n);
        for ( ; min_i < i; ++min_i)
          result[perm[min_i]] = res;
      }
    }


    // weighted version
    template<typename RandomAccessIter1, typename RandomAccessIter2>
    void normalize(RandomAccessIter1 first, RandomAccessIter1 last,
                   RandomAccessIter2 result, 
                   utility::weighted_iterator_tag) const
    {
      std::copy(utility::weight_iterator(first), 
                utility::weight_iterator(last),
                utility::weight_iterator(result));
      std::vector<double> input_vec;
      input_vec.reserve(last-first);
      // set values with w=0 to 0 to avoid problems with NaNs
      utility::iterator_traits<RandomAccessIter1> trait;
      for (RandomAccessIter1 i=first; i!=last; ++i) {
        if (trait.weight(i)==0)
          input_vec.push_back(0.0);
        else
          input_vec.push_back(trait.data(i));
      }

      std::vector<size_t> perm(input_vec.size());
      utility::sort_index(input_vec.begin(), input_vec.end(), perm);
      utility::iterator_traits<RandomAccessIter2> rtrait;

      double sum_w=0;
      size_t i=0; 
      while ( i<perm.size() ) {
        double w=0;
        size_t min_i = i;
        while (i<perm.size() && (trait.weight(first+perm[i])==0 ||
                                 trait.data(first+perm[i]) <= 
                                 trait.data(first+perm[min_i])) ) {
          w += trait.weight(first+perm[i]);
          ++i;
        }
        double res=sum_w + 0.5*w;
        for ( size_t j=min_i; j<i; ++j)
          rtrait.data(result+perm[j]) = res;
        sum_w += w;
      }       

      size_t n = std::distance(first, last);
      std::transform(utility::data_iterator(result),
                     utility::data_iterator(result+n),
                     utility::data_iterator(result),
                     std::bind2nd(std::divides<double>(), sum_w));
    }

  };

}}} // end of namespace normalizer, yat and thep
#endif