source: trunk/src/InputRanker.cc @ 89

// $Id: InputRanker.cc 89 2004-05-28 17:44:42Z peter $

// Thep C++ Tools
#include "InputRanker.h"
#include "ROC.h"
#include "matrix.h"
#include "vector.h"
#include "stl_utility.h"

#include <vector>
#include <utility>

namespace theplu {
namespace cpptools {  

InputRanker::InputRanker(const gslapi::matrix& data, 
                         const gslapi::vector& target, 
                         const std::vector<size_t> train_set)
  :train_set_(train_set),
   id_(std::vector<size_t>(data.rows())),
   rank_(std::vector<size_t>(data.rows()))
 
{
  using namespace std;
  // Peter, this copying and transposing is done since it seems
  // problematic to get a column vector from a matrix
  gslapi::matrix data_transposed = data;
  data_transposed.transpose();
  
  size_t nof_genes = data_transposed.rows();
  size_t nof_samples = data_transposed.columns();
  if (!train_set_.size())
    for (size_t i=0; i<nof_samples; i++)
      train_set_.push_back(i);  

  //scoring each gene
  std::vector<pair<size_t, double> > score;
  for (size_t i=0; i<nof_genes; i++){
    ROC roc(data_transposed[train_set_[i]], target);
    std::pair<size_t, double> tmp(i,roc.area());
    score.push_back(tmp);
  }
  //sort the scores and assign id_ and rank_
  sort(score.begin(), score.end(), 
       pair_value_compare<size_t, double>());
  
  for (size_t i=0; i<nof_genes; i++){
    id_[i]=score[i].first;
    rank_[id_[i]]=i;
  }
  
  
}

}} // of namespace cpptools and namespace theplu