STK_GammaBase.h

Go to the documentation of this file.

/*--------------------------------------------------------------------*/
/*     Copyright (C) 2004-2016 Serge Iovleff
 
    This program 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 2 of the
    License, or (at your option) any later version.
 
    This program 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 this program; if not, write to the
    Free Software Foundation, Inc.,
    59 Temple Place,
    Suite 330,
    Boston, MA 02111-1307
    USA
 
    Contact : S..._DOT_I..._AT_stkpp.org (see copyright for ...)
*/
 
/*
 * Project:  stkpp::Clustering
 * created on: Dec 4, 2013
 * Authors: Serge Iovleff
 **/
 
#ifndef STK_GAMMABASE_H
#define STK_GAMMABASE_H
 
#include "../STK_IMixtureDensity.h"
 
#include <STatistiK/include/STK_Law_Gamma.h>
#include <STatistiK/include/STK_Stat_Functors.h>
 
#include <Analysis/include/STK_Algo_FindZero.h>
#include <Analysis/include/STK_Funct_raw.h>
#include <Analysis/include/STK_Funct_gamma.h>
#include "../GammaModels/STK_GammaParameters.h"
 
namespace STK
{
 
namespace hidden
{
class invPsiMLog: public IFunction<invPsiMLog >
{
  public:
    inline invPsiMLog( Real const& y): y_(y)  {}
    inline Real fImpl(Real const& a) const
    { return (y_ + std::log(a) - Funct::psi_raw(a));}
    inline Real xminImpl() const { return 0;}
 
  private:
    Real y_;
};
 
class invPsi: public IFunction<invPsi >
{
  public:
    inline invPsi( Real const& y): y_(y) {}
    inline Real fImpl(Real const& x) const { return (y_ - Funct::psi_raw(x));}
    inline Real xminImpl() const { return 0;}
  private:
    Real y_;
};
 
} // namespace hidden
 
template<class Derived>
class GammaBase: public IMixtureDensity<Derived >
{
  public:
    typedef IMixtureDensity<Derived > Base;
    using Base::param_;
    using Base::p_data;
 
  protected:
    GammaBase( int nbCluster): Base(nbCluster) {}
    GammaBase( GammaBase const& model): Base(model) {}
    ~GammaBase() {}
 
  public:
    inline Real shape(int k, int j) const { return param_.shape(k,j);}
    inline Real scale(int k, int j) const { return param_.scale(k,j);}
 
    void initializeModelImpl() { param_.resize(p_data()->cols());}
 
    Real lnComponentProbability(int i, int k) const;
    template<class Weights>
    Real impute(int i, int j, Weights const& pk) const;
    inline Real rand(int i, int j, int k) const
    { return Law::Gamma::rand(shape(k,j), scale(k,j));}
    template<class Array>
    void getParameters(Array& params) const;
    void writeParameters(CArrayXX const* p_tik, ostream& os) const;
 
  protected:
    Real qValue(CArrayXX const* p_tik, CPointX const* p_tk) const;
    bool moments(CArrayXX const* p_tik);
    inline Real meanjk( int j, int k) { return param_.mean_[k][j];}
    inline Real variancejk( int j, int k) { return param_.variance_[k][j];}
    inline Real meank( int k) { return param_.mean_[k].mean();}
    inline Real variancek( int k) { return param_.variance_[k].mean();}
};
 
template<class Derived>
Real GammaBase<Derived>::lnComponentProbability(int i, int k) const
{
  Real sum =0.;
  for (int j=p_data()->beginCols(); j<p_data()->endCols(); ++j)
  {
    if (param_.shape(k,j) && param_.scale(k,j))
    { sum += Law::Gamma::lpdf(p_data()->elt(i,j), param_.shape(k,j), param_.scale(k,j));}
  }
  return sum;
}
/* @return an imputation value for the jth variable of the ith sample
 *  @param i,j indexes of the data to impute
 *  @param pk the probabilities of each class for the ith individual
 **/
template<class Derived>
template<class Weights>
Real GammaBase<Derived>::impute(int i, int j, Weights const& pk) const
{
  Real sum = 0.;
  for (int k= pk.begin(); k < pk.end(); ++k)
  { sum += pk[k] * shape(k,j) * scale(k,j);}
  return sum;
}
 
/* compute safely the weighted moments of a gamma law. */
template<class Derived>
bool GammaBase<Derived>::moments(CArrayXX const* p_tik)
{
  for (int k= p_tik->beginCols(); k < p_tik->endCols(); ++k)
  {
    CVectorX tikColk(p_tik->col(k), true); // create a reference
    for (int j=p_data()->beginCols(); j<p_data()->endCols(); ++j)
    {
      // mean
      Real mean =  p_data()->col(j).wmean(tikColk);
      if ( (mean<=0) || isNA(mean) ) { return false;}
      param_.mean_[k][j] = mean;
      // mean log
      Real meanLog =  p_data()->col(j).log().wmean(tikColk);
      if (isNA(meanLog)) { return false;}
      param_.meanLog_[k][j] = meanLog;
      // variance
      Real variance =  p_data()->col(j).wvariance(mean, tikColk);
      if ((variance<=0)||isNA(variance)){ return false;}
      param_.variance_[k][j] = variance;
    }
  }
  return true;
}
 
/* compute the intermediate value needed by Newton algorithm*/
template<class Derived>
Real GammaBase<Derived>::qValue(CArrayXX const* p_tik, CPointX const* p_tk) const
{
  Real value = 0.;
  for (int k= p_tik->beginCols(); k < p_tik->endCols(); ++k)
  {
    Real sumjk=0.0;
    for (int j=p_data()->beginCols(); j<p_data()->endCols(); ++j)
    {
      Real a = shape(k,j), b = scale(k,j);
      sumjk += a * (param_.meanLog_[k][j]-std::log(b))
             - param_.mean_[k][j]/b - STK::Funct::lgamma(a);
    }
    value += p_tk->elt(k) * sumjk;
  }
  return value;
}
 
template<class Derived>
void GammaBase<Derived>::writeParameters(CArrayXX const* p_tik, ostream& os) const
{
  CPointX a(p_data()->cols()), b(p_data()->cols());
  for (int k= p_tik->beginCols(); k < p_tik->endCols(); ++k)
  {
    // store shape and scale values in an array for a nice output
    for (int j=p_data()->beginCols();  j < p_data()->endCols(); ++j)
    {
      a[j] = shape(k,j);
      b[j] = scale(k,j);
    }
    os << _T("---> Component ") << k << _T("\n");
    os << _T("shape = ") << a;
    os << _T("scale = ") << b;
  }
}
 
template<class Derived>
template<class Array>
void GammaBase<Derived>::getParameters(Array& params) const
{
  int nbClust = this->nbCluster();
  params.resize(2*nbClust, p_data()->cols());
  for (int k= 0; k < nbClust; ++k)
  {
    for (int j=  p_data()->beginCols();  j< p_data()->endCols(); ++j)
    {
      params(baseIdx+2*k  , j) = shape(baseIdx+k,j);
      params(baseIdx+2*k+1, j) = scale(baseIdx+k,j);
    }
  }
}
 
 
 
} // namespace STK
 
#endif /* STK_GammaBASE_H */