PreChiSquare.hh

Go to the documentation of this file.

#ifndef PRE_CHISQUARE_HH
#define PRE_CHISQUARE_HH

#include "PreAlgorithm.hh"

#include <openmodeller/om.hh>
#include <openmodeller/Exceptions.hh>
#include <openmodeller/Sample.hh>
#include <openmodeller/Sampler.hh>

#define classLimit 16

/*
 *
 *
 * input: SamplerPtr output: shows for each variable the amount of layers that
 * maintain correlation at the significance level of 0.05
 */

class dllexp PreChiSquare : public PreAlgorithm 
{
  private:
    size_t  num_points;     // number of points.
    size_t  num_layers;     // number of layers.
    size_t  nclass;   // number of class
    Sample  minimum;  // Mininum value for each layer.
    Sample  delta;    // delta value for each layer. 
    OccurrencesPtr my_presences; //occurrence points of species.
    Scalar  measured[classLimit][classLimit]; //measured points between two crossed layers.
    Scalar  expected [classLimit][classLimit]; //expected points between two crossed layers.
    Scalar    chicell [classLimit][classLimit]; //chi value for each cell formed between two class of the crossed layers. 
    std::vector< size_t > statistic1; //dependency between layers.
    std::vector< size_t > statistic2; //dependency between layers.

  public:

    PreChiSquare(); //constructor
    ~PreChiSquare(); //destructor

    //Return description about the algorithm
    string getDescription() const { return "\
This technique is described in Li, L., et al. (2006). \"An Integrated\
Bayesian Modelling Approach for Predicting Mosquito Larval Habitats\". The \
authors tested the independence between all pairs of layers. Each pair is \
cross-referenced using a contingency matrix to detect the dependency between \
the layers based on a set of presence points from a particular species. \
The output shows for each layer the number of layers that maintain correlation \
with the significance level of 0.05."; }

    //Checks if the supplied parameters fits the requirements of PRE algorithm implementation.
    //return true if the parameters are OK. false if not.
    bool checkParameters( const PreParameters& parameters ) const;

    //Runs the current algorithm implementation.
    //return true if OK. false on error.
    bool runImplementation();

    //get input parameters
    void getAcceptedParameters ( stringMap& info );

    //get output information
    void getLayersetResultSpec ( stringMap& info);

    //get output information for each layer
    void getLayerResultSpec ( stringMap& info);

    //initialize attributes: num_points, num_layers, nclass, minimum, delta, my_presences.
    bool init();

    //get number of points.
    size_t getNpoints();

    //set number of points.
    void setNpoints();

    //get number of layers.
    size_t getNlayers();

    //set number of layers. 
    void setNlayers(SamplerPtr samplerPtr);

    //set number of class.
    size_t getNclass();

    //set number of class.
    void setNclass();

    //set mininum value for each layer.
    void setMinimum();

    //get minimum value of each layer.
    Sample getMinimum();

    //set delta value for each layer. 
    void setDelta();

    //get delta value of each layer. 
    Sample getDelta();

    //set measured points for crossed layers.
    void setMeasured(size_t layer1, size_t layer2);

    //set expected points for crossed layers.
    void setExpected();

    //set chi value for each cell formed between two class of the crossed layers. 
    void setChicell();

    //count layer dependency.
    void setStatistic(size_t layer1, size_t layer2);

    //control layers crossing.
    void run();

    //set zero to all elements of the matrix "measured".
    void setMeasuredZero();

};


#endif