org.metaqtl
Class EMResult

java.lang.Object
  extended by org.metaqtl.EMResult

public class EMResult
extends java.lang.Object


Field Summary
 double[] ccov
          The complete information (just for means).
 double clog
          The complete loglikelihood.
 EMCriteria criteria
          The model choice criteria.
 double[][] dm
          The matrix of convergence rate for the EM-Algorithm (just for means).
 double edist
          The euclidean distance between the current and the previous vector of parameters (mu + pi)
 int k
          The number of clusters.
 double[] mu
          The means of the clusters.
 int n
          The number of data points.
 double[][] ocov
          The observed information (just for means).
 double olog
          The observed loglikelihood.
 double[] pi
          The mixing proportions.
 double rate
          The quadratric rate of convergence of the algorithm.
 double[][] z
          The cluster membership probabilties.
 
Constructor Summary
EMResult()
           
EMResult(int n, int k)
           
 
Method Summary
 void computeCriteria()
          This methos computes the following criteria : AIC AIC3 ICOMP MIR BIC AWE
static void copy(EMResult dest, EMResult src)
           
 double getCriterion(int critIdx)
           
 double[] getEuclidean()
           
 int getK()
           
 double[] getMahalanobis(double[] sd)
           
 double getMu(int i)
           
 double getPi(int i)
           
 double getSD(int i)
           
 double[] getXBestPred()
           
 double[] getXPred()
           
 int getXPredIdx(int i)
           
 double[] getZ(int i)
           
 void sortCluster()
          Sort the components in increasing order.
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Field Detail

k

public int k
The number of clusters.


n

public int n
The number of data points.


rate

public double rate
The quadratric rate of convergence of the algorithm.


edist

public double edist
The euclidean distance between the current and the previous vector of parameters (mu + pi)


mu

public double[] mu
The means of the clusters.


pi

public double[] pi
The mixing proportions.


z

public double[][] z
The cluster membership probabilties.


olog

public double olog
The observed loglikelihood.


clog

public double clog
The complete loglikelihood.


dm

public double[][] dm
The matrix of convergence rate for the EM-Algorithm (just for means).


ocov

public double[][] ocov
The observed information (just for means).


ccov

public double[] ccov
The complete information (just for means).


criteria

public EMCriteria criteria
The model choice criteria.

Constructor Detail

EMResult

public EMResult()

EMResult

public EMResult(int n,
                int k)
Parameters:
n - the number of data points.
k - the number of components.
Method Detail

sortCluster

public void sortCluster()
Sort the components in increasing order.


copy

public static void copy(EMResult dest,
                        EMResult src)
Parameters:
theta -
spoint -

computeCriteria

public void computeCriteria()
This methos computes the following criteria :


getCriterion

public double getCriterion(int critIdx)
Parameters:
critIdx -

getMahalanobis

public double[] getMahalanobis(double[] sd)
Parameters:
sd -

getXPred

public double[] getXPred()
Returns:

getSD

public double getSD(int i)
Parameters:
i -
Returns:

getMu

public double getMu(int i)
Parameters:
i -
Returns:

getK

public int getK()
Returns:

getZ

public double[] getZ(int i)

getEuclidean

public double[] getEuclidean()
Parameters:
sd -
Returns:

getPi

public double getPi(int i)
Parameters:
i -
Returns:

getXPredIdx

public int getXPredIdx(int i)
Parameters:
i -
Returns:

getXBestPred

public double[] getXBestPred()
Returns: