grid_cluster_analysis.cpp

这是一个GPS相关的程序

///////////////////////////////////////////////////////////
//                                                       //
//                         SAGA                          //
//                                                       //
//      System for Automated Geoscientific Analyses      //
//                                                       //
//                    Module Library:                    //
//                  Grid_Discretisation                  //
//                                                       //
//-------------------------------------------------------//
//                                                       //
//               Grid_Cluster_Analysis.cpp               //
//                                                       //
//                 Copyright (C) 2003 by                 //
//                      Olaf Conrad                      //
//                                                       //
//-------------------------------------------------------//
//                                                       //
// This file is part of 'SAGA - System for Automated     //
// Geoscientific Analyses'. SAGA 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; version 2 of the License.   //
//                                                       //
// SAGA 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.                                                  //
//                                                       //
//-------------------------------------------------------//
//                                                       //
//    e-mail:     [e-mail]                   //
//                                                       //
//    contact:    Olaf Conrad                            //
//                Institute of Geography                 //
//                University of Goettingen               //
//                Goldschmidtstr. 5                      //
//                37077 Goettingen                       //
//                Germany                                //
//                                                       //
///////////////////////////////////////////////////////////

//---------------------------------------------------------


///////////////////////////////////////////////////////////
//														 //
//														 //
//														 //
///////////////////////////////////////////////////////////

//---------------------------------------------------------
#include "Grid_Cluster_Analysis.h"


///////////////////////////////////////////////////////////
//														 //
//														 //
//														 //
///////////////////////////////////////////////////////////

//---------------------------------------------------------
CGrid_Cluster_Analysis::CGrid_Cluster_Analysis(void)
{
	//-----------------------------------------------------
	// 1. Info...

	Set_Name		(_TL("Cluster Analysis for Grids"));

	Set_Author		(_TL("Copyrights (c) 2001 by Olaf Conrad"));

	Set_Description	(_TW(		
		"Cluster Analysis for grids.\n\nReferences:\n\n"

		"Iterative Minimum Distance:\n"
		"- Forgy, E. (1965):\n"
		"  'Cluster Analysis of multivariate data: efficiency vs. interpretability of classifications',\n"
		"  Biometrics 21:768\n\n"

		"Hill-Climbing:"
		"- Rubin, J. (1967):\n"
		"  'Optimal Classification into Groups: An Approach for Solving the Taxonomy Problem',\n"
		"  J. Theoretical Biology, 15:103-144\n\n"
	));


	//-----------------------------------------------------
	// 2. Grids...

	Parameters.Add_Grid_List(
		NULL	, "INPUT"		, _TL("Grids"),
		_TL(""),
		PARAMETER_INPUT
	);

	Parameters.Add_Grid(
		NULL	, "RESULT"		, _TL("Clusters"),
		_TL(""),
		PARAMETER_OUTPUT, true, GRID_TYPE_Int
	);

	Parameters.Add_Table(
		NULL	, "STATISTICS"	, _TL("Statistics"),
		_TL(""),
		PARAMETER_OUTPUT
	);


	//-----------------------------------------------------
	// 3. General Parameters...

	Parameters.Add_Choice(
		NULL	, "METHOD"		, _TL("Method"),
		_TL(""),
		CSG_String::Format(SG_T("%s|%s|%s|"),
			_TL("Iterative Minimum Distance (Forgy 1965)"),
			_TL("Hill-Climbing (Rubin 1967)"),
			_TL("Combined Minimum Distance / Hillclimbing") 
		),1
	);

	Parameters.Add_Value(
		NULL	, "NCLUSTER"	, _TL("Clusters"),
		_TL("Number of clusters"),
		PARAMETER_TYPE_Int, 10, 2, true
	);

	Parameters.Add_Value(
		NULL	, "NORMALISE"	, _TL("Normalise"),
		_TL("Automatically normalise grids by standard deviation before clustering."),
		PARAMETER_TYPE_Bool, true
	);

	Parameters.Add_Value(
		NULL	, "UPDATEVIEW"	, _TL("Update View"),
		_TL("Update cluster view while clustering."),
		PARAMETER_TYPE_Bool, true
	);
}

//---------------------------------------------------------
CGrid_Cluster_Analysis::~CGrid_Cluster_Analysis(void)
{}


///////////////////////////////////////////////////////////
//														 //
//														 //
//														 //
///////////////////////////////////////////////////////////

//---------------------------------------------------------
bool CGrid_Cluster_Analysis::On_Execute(void)
{
	int						i, j, nCluster;
	long					nElements;
	double					SP;
	CSG_Parameter_Grid_List	*pGrids;

	//-----------------------------------------------------
	pGrids		= Parameters("INPUT")	->asGridList();
	nGrids		= pGrids->Get_Count();

	pCluster	= Parameters("RESULT")	->asGrid();
	nCluster	= Parameters("NCLUSTER")->asInt();

	//-----------------------------------------------------
	if( nGrids > 0 )
	{
		Grids		= (CSG_Grid **)SG_Malloc(nGrids * sizeof(CSG_Grid *));

		if( Parameters("NORMALISE")->asBool() )
		{
			for(i=0; i<nGrids; i++)
			{
				Grids[i]	= SG_Create_Grid(pGrids->asGrid(i), GRID_TYPE_Float);
				Grids[i]->Assign(pGrids->asGrid(i));
				Grids[i]->Normalise();
			}
		}
		else
		{
			for(i=0; i<nGrids; i++)
			{
				Grids[i]	= pGrids->asGrid(i);
			}
		}

		pCluster->Assign(-1);

		nMembers	= (int     *)SG_Malloc(nCluster * sizeof(int));
		Variances	= (double  *)SG_Malloc(nCluster * sizeof(double));
		Centroids	= (double **)SG_Malloc(nCluster * sizeof(double *));

		for(i=0; i<nCluster; i++)
		{
			Centroids[i]	= (double  *)SG_Malloc(nGrids * sizeof(double));
		}

		//-------------------------------------------------
		nElements	= Get_NCells();

		switch( Parameters("METHOD")->asInt() )
		{
		case 0:
			SP	= MinimumDistance	(nElements, nCluster);
			break;

		case 1:
			SP	= HillClimbing		(nElements, nCluster);
			break;

		case 2:
			SP	= MinimumDistance	(nElements, nCluster);

			nElements	= Get_NCells();	// may have been diminished because of no data values...

			SP	= HillClimbing		(nElements, nCluster);
			break;
		}

		//-------------------------------------------------
		if( Parameters("NORMALISE")->asBool() )
		{
			for(i=0; i<nGrids; i++)
			{
				delete(Grids[i]);

				Grids[i]	= pGrids->asGrid(i);
			}

			for(i=0; i<nGrids; i++)
			{
				for(j=0; j<nCluster; j++)
				{
					Centroids[j][i]	= sqrt(Grids[i]->Get_Variance()) * Centroids[j][i] + Grids[i]->Get_ArithMean();
				}
			}
		}

		Write_Result(Parameters("STATISTICS")->asTable(), nElements, nCluster, SP);

		SG_Free(Grids);

		for(i=0; i<nCluster; i++)
		{
			SG_Free(Centroids[i]);
		}

		SG_Free(Centroids);
		SG_Free(Variances);
		SG_Free(nMembers);

		return( true );
	}

	return( false );
}


///////////////////////////////////////////////////////////
//														 //
//														 //
//														 //
///////////////////////////////////////////////////////////

//---------------------------------------------------------
void CGrid_Cluster_Analysis::Write_Result(CSG_Table *pTable, long nElements, int nCluster, double SP)
{
	int				i, j;
	CSG_String		s;
	CSG_Table_Record	*pRecord;

	pTable->Destroy();
	pTable->Set_Name(_TL("Cluster Analysis"));

	pTable->Add_Field(_TL("ClusterID")	, TABLE_FIELDTYPE_Int);
	pTable->Add_Field(_TL("Elements")	, TABLE_FIELDTYPE_Int);
	pTable->Add_Field(_TL("Variance")	, TABLE_FIELDTYPE_Double);

	s.Printf(SG_T("\n%s:\t%ld \n%s:\t%d \n%s:\t%d \n%s:\t%f"),
		_TL("Number of Elements")			, nElements,
		_TL("\nNumber of Variables")		, nGrids,
		_TL("\nNumber of Clusters")			, nCluster,
		_TL("\nValue of Target Function")	, SP
	);

	s.Append(CSG_String::Format(SG_T("%s\t%s\t%s"), _TL("Cluster"), _TL("Elements"), _TL("Variance")));

	for(j=0; j<nGrids; j++)
	{
		s.Append(CSG_String::Format(SG_T("\t%02d_%s"), j + 1, Grids[j]->Get_Name()));
		pTable->Add_Field(Grids[j]->Get_Name(), TABLE_FIELDTYPE_Double);
	}

	Message_Add(s);

	for(i=0; i<nCluster; i++)
	{
		s.Printf(SG_T("%d\t%d\t%f"), i, nMembers[i], Variances[i]);

		pRecord	= pTable->Add_Record();
		pRecord->Set_Value(0, i);
		pRecord->Set_Value(1, nMembers[i]);
		pRecord->Set_Value(2, Variances[i]);

		for(j=0; j<nGrids; j++)
		{
			s.Append(CSG_String::Format(SG_T("\t%f"), Centroids[i][j]));

			pRecord->Set_Value(j + 3, Centroids[i][j]);
		}

		Message_Add(s);
	}
}


///////////////////////////////////////////////////////////
//														 //
//					Minimum Distance					 //
//														 //
///////////////////////////////////////////////////////////

//---------------------------------------------------------
// (nElements)-Array -> Bei Eingabe Startgruppierung oder 0.
// Bei Ausgabe Gruppierung: >Das ite Element ist im Cluster Cluster(i).

double CGrid_Cluster_Analysis::MinimumDistance(long &nElements, int nCluster)
{
	//-----------------------------------------------------
	// Variablen...

	bool	bContinue;
	int		iElement, iGrid, iCluster, nClusterElements, nShifts, minCluster, nPasses;
	double	d, Variance, minVariance, SP, SP_Last	= -1;


	//-----------------------------------------------------
	// Anfangspartition (Standard falls nicht vorgegeben

	for(iElement=0, nClusterElements=0; iElement<nElements; iElement++)
	{
		for(iGrid=0, bContinue=true; iGrid<nGrids && bContinue; iGrid++)
		{
			if( Grids[iGrid]->is_NoData(iElement) )
			{
				bContinue	= false;
			}
		}

		if( bContinue )
		{