scalablekmeans.h

Scalable k-means software and test datasets This package (a Unix tar file, gzipped) contains the sou

/* Scalable K-means clustering software
Copyright (C) 2000  Fredrik Farnstrom and James Lewis

This program 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 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 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.


See the file README.TXT for more information.

*/

/* scalablekmeans.h */

class Database;
class Subcluster;
class MainCluster;
class Singleton;
class SyntheticData;

struct ClusterNode
{
	float distance;
	Singleton *point;
	static int compareDistance(const void *a, const void *b);
};

class ScalableKMeans
{
	RAMBuffer buffer;
	Database *database;
	int numClusters;

	float beta;
	float convergenceThreshold;
	float secondaryConvergenceThreshold;
	float discardFraction;
	float confidenceStdDev;
	int numNewSecondaryClusters;

	long numMainIterations;
	long numSecondaryIterations;
	long numUpdates;

	int primary1:1, primary2:1, secondary:1, naive:1;

	static const float defaultBeta = 1.1;
	static const float defaultConvergenceThreshold = 0.0001;
	static const float defaultSecondaryCT = 0.0001;
	static const float defaultDiscardFraction = 0.2;
	static const float defaultConfidenceStdDev = 3.4;
//	static const float defaultConfidenceStdDev = 3.7;
//	static const float defaultConfidenceStdDev = 3.1;

	MainCluster *mainClusters;
	Subcluster *oldClusters;

	void setup(void);
	void initModel(void);

	int fillBuffer(void);

	void naiveCompression(void);
	void primaryCompression(void);
	void secondaryCompression(void);

	void exportCompressedSet(void);

	int updateModel(void);

	Subcluster *closestMainCluster(Singleton *p);
	Subcluster *closestMainCluster(Subcluster *c);

	double permutationDistance(int *permut, SyntheticData *db);

	FILE *logFile;

public:
	ScalableKMeans(long bufferSize, Database *db, int k, int dim);

	~ScalableKMeans();

	void cluster(void);

	double logLikelihood(void);
	double distanceToTrueClusters(SyntheticData *db);

	void lesion(int p1, int p2, int s, int n)
	{	primary1 = p1;
		primary2 = p2;
		secondary = s;
		naive = n;
	}

	float mainIterations(void)
	{	return ((float)numMainIterations)/numUpdates;
	}

	float secondaryIterations(void)
	{	return ((float)numSecondaryIterations)/numUpdates;
	}

	void setConstants(float newbeta, float newct, float newsct, float newdf,
			float newcsd)
	{	beta = newbeta;
		convergenceThreshold = newct;
		secondaryConvergenceThreshold = newsct;
		discardFraction = newdf;
		confidenceStdDev = newcsd;
	}

	long getNumPoints(void)
	{	return buffer.getTotalPoints();
	}

	void logUpdates(char *filename);
};


/* End of file scalablekmeans.h */