jminhep.cluster
Class Partition

java.lang.Object
  extended by jminhep.cluster.Partition

public class Partition
extends java.lang.Object

Interface class to all clustering algorithms. Use it for JMInHEP embeded applications


Constructor Summary
Partition(DataHolder data)
           
 
Method Summary
 DataHolder getCenters()
          Get cluster centers
 double getCompactness()
          Get compactness of the cluster configurations
 java.lang.String getName()
          Get name for this clustering
 int getNclusters()
          get number of reconstructed clusters
 int[] getPoints()
          Get number of points accosiated with clusters
 DataHolder getSeeds()
          Get seeds
 void run(int Imode)
          Main method to run cluster algorithm
 void set(int Nclusters, double Eps, double Fuzzines, int Niterations)
          Set parameters for clustering
 void setProbab(double probability)
          Set probability for cluster accosiation
 
Methods inherited from class java.lang.Object
equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Constructor Detail

Partition

public Partition(DataHolder data)
Method Detail

set

public void set(int Nclusters,
                double Eps,
                double Fuzzines,
                int Niterations)
Set parameters for clustering

Parameters:
Nclusters - - number of clusters
Eps - - precision of clustering
Fuzzines - - fuzziness (only for cmeans algorithms, for others just dummy constant)
Niterations - - Max number of iterations

run

public void run(int Imode)
Main method to run cluster algorithm

Parameters:
Imode - - cluster mode (int). Find the correct mode from the example or by running JMinHEP in GUI mode. After each clustering, Imode value is shown in the status bar.

getName

public java.lang.String getName()
Get name for this clustering

Returns:
Description of the cluster mode

getCompactness

public double getCompactness()
Get compactness of the cluster configurations

Returns:
compactness

getCenters

public DataHolder getCenters()
Get cluster centers

Returns:
DataHolder with cluster centers

getSeeds

public DataHolder getSeeds()
Get seeds

Returns:
DataHolder with seed values (only for k-means algorithms)

getPoints

public int[] getPoints()
Get number of points accosiated with clusters

Returns:
number of points in each cluster (int[])

setProbab

public void setProbab(double probability)
Set probability for cluster accosiation

Parameters:
probability - Active only for cmeans (and only for representative purpose)

getNclusters

public int getNclusters()
get number of reconstructed clusters

Returns:
number of reconstructed clusters


jHepWork 1.1 (C) Chekanov